Glossary of Terms
Activity Types
Categories used to classify distinct kinds of employee actions recorded as events in an organizational graph, such as communication, collaboration, learning, or administrative work.
Example: A graph model defines activity types including "email sent," "meeting attended," "document authored," and "training completed" to categorize event nodes.
AI-generated Content
Content produced by artificial intelligence systems, including text, images, code, or structured data, that may appear within organizational event streams and requires identification to maintain data provenance and trust.
Example: A chatbot automatically generates summary notes after each team meeting and inserts them into the collaboration platform's activity log.
Alerting Systems
Automated mechanisms that monitor organizational data streams and deliver notifications when predefined thresholds are crossed or significant events are detected, enabling timely response to emerging conditions.
Example: A system sends an alert to HR when a department's voluntary turnover rate exceeds 15% within a rolling quarter.
Alignment Analysis
The process of measuring the degree to which observed organizational activities, resource allocations, and communication patterns correspond to declared strategic objectives.
Example: Graph queries reveal that 40% of cross-team collaboration edges connect to projects not linked to any current strategic initiative.
Anomaly Detection
The identification of data points, patterns, or events that deviate significantly from established baselines or expected behavior within organizational data, signaling potential issues or opportunities.
Example: A sudden spike in after-hours email volume from a single department triggers an anomaly flag in the monitoring dashboard.
Anonymization
The irreversible process of removing or transforming personally identifiable information from a dataset so that the data subject can no longer be identified, directly or indirectly.
Example: Replacing employee names and IDs with random tokens and stripping metadata so that communication graph patterns can be analyzed without revealing individual identities.
API Integration
The process of connecting disparate organizational systems through standardized programming interfaces to enable automated data exchange, event capture, and coordinated functionality across platforms.
Example: An HR information system exposes a REST API that feeds employee role-change events directly into the organizational graph database.
Audit Trails
Chronological records that capture the sequence of system activities, including who accessed or modified data, what changes were made, and when they occurred.
Example: A graph database logs every query against employee nodes, recording the analyst's identity, timestamp, and the specific properties accessed.
Average Path Length
The mean number of edges along the shortest paths between all reachable pairs of nodes in a graph, indicating how quickly information or influence can traverse the network.
Example: An organization with an average path length of 3.2 means any two employees are, on average, about three introductions apart.
Backlog Task Assignment
The process of allocating queued, unassigned work items to available personnel using criteria such as skill match, current workload, priority, and deadline proximity.
Example: A graph query identifies three analysts with matching skills and below-average task loads to assign pending data quality reviews.
Batch Loading
A data ingestion method that collects records over a defined interval and loads them into the target database as a single scheduled operation rather than individually upon arrival.
Example: An overnight job loads all new employee onboarding records from the past 24 hours into the graph database at 2:00 AM.
Benchmarking
The practice of comparing an organization's metrics, processes, or performance indicators against established standards, industry norms, or peer organizations to identify relative strengths and gaps.
Example: A company compares its average onboarding-to-productivity time of 90 days against an industry median of 60 days.
Betweenness Centrality
A graph metric that quantifies a node's importance by calculating the proportion of shortest paths between all other node pairs that pass through it, identifying bridging roles.
Example: An employee with high betweenness centrality sits between two departments, serving as the primary conduit for cross-team information flow.
Bias in Analytics
Systematic distortion in analytical outputs caused by prejudice, unrepresentative data, or flawed algorithmic assumptions that produces unfair or inaccurate results for certain groups.
Example: A promotion-readiness model trained predominantly on data from one demographic group may systematically undervalue candidates from underrepresented groups.
Boundary Spanners
Individuals who maintain active connections across formal organizational boundaries such as departments, divisions, or geographic locations, enabling knowledge transfer and coordination between otherwise separated units.
Example: A product manager who regularly collaborates with engineering, marketing, and customer support serves as a boundary spanner linking three distinct functional areas.
Breadth-first Search
A graph traversal algorithm that explores all nodes at the current depth level before moving to nodes at the next depth level, proceeding outward from a starting node layer by layer.
Example: Starting from a CEO node, BFS first visits all direct reports, then all of their direct reports, mapping the organization level by level.
Bridge Builders
Individuals whose relationships connect otherwise disconnected or weakly connected groups within a network, facilitating communication and collaboration across structural gaps.
Example: A senior analyst who participates in both the data science team's and the finance team's meetings bridges two groups that rarely interact directly.
Building a Graph Library
The process of creating, curating, and maintaining a collection of reusable graph database components, including queries, algorithms, data models, and visualization templates for organizational analytics.
Business Process Mining
An analytical discipline that applies algorithms to event log data to discover, monitor, and improve actual business processes as they are executed within an organization.
Example: Analyzing timestamped approval events reveals that purchase orders follow six distinct paths rather than the single path documented in the policy manual.
Calendar Events
Structured records generated by scheduling systems that capture the creation, modification, attendance, and completion of time-bound appointments and meetings.
Example: A calendar event record includes the organizer, invitees, start time, duration, recurrence pattern, and room booking for a weekly team standup.
Career Guidance
Personalized recommendations for professional development and progression, derived from analysis of skill profiles, organizational needs, historical career trajectories, and available opportunities within the network.
Example: The system recommends that an analyst pursue a project management certification based on successful paths taken by similar employees.
Career Path Analysis
The examination of historical and potential progression routes within an organization, using graph traversal of role transitions, skill acquisitions, and promotion patterns to reveal viable advancement opportunities.
Example: A path query shows that employees who moved from data analyst to project lead most frequently passed through a business analyst role first.
Chat Event Streams
Continuous sequences of timestamped records produced by instant messaging and collaboration platforms, capturing message sends, channel joins, reactions, and file shares.
Example: A Slack workspace generates chat events each time a user posts a message, threads a reply, or adds an emoji reaction in a channel.
Closeness Centrality
A graph metric that measures a node's importance by computing the reciprocal of the average shortest-path distance from that node to all other reachable nodes in the graph.
Example: An employee with high closeness centrality can reach anyone in the organization through fewer communication hops, making them efficient information disseminators.
Clustering Coefficient
A metric that measures the degree to which a node's neighbors are connected to each other, expressed as the ratio of existing edges among neighbors to the maximum possible edges.
Example: A clustering coefficient of 0.8 for an employee node means that 80% of that person's contacts also communicate directly with one another.
Communication Bottlenecks
Points in an organizational network where information flow is disproportionately constrained, typically because a small number of nodes or edges carry traffic that exceeds their capacity.
Example: If all project updates must pass through a single team lead before reaching executives, that team lead is a communication bottleneck.
Communication Channels
The distinct pathways or media through which organizational members exchange information, represented as edge types or properties in a communication graph.
Example: A graph model represents email, Slack messages, video calls, and in-person meetings as separate relationship types between employee nodes.
Communication Frequency
A quantitative property on communication edges that records how often interactions occur between two nodes within a defined time period.
Example: An edge between two employee nodes carries a frequency property of 12, indicating twelve email exchanges during the past month.
Communication Tone Analysis
The systematic assessment of the style, manner, and register of written or spoken communications within an organization, used to evaluate professionalism, collaboration patterns, and interpersonal dynamics.
Communication Volume
A quantitative measure of the total amount of communication between nodes, typically captured as an edge property representing message count, word count, or duration.
Example: The volume property on a communication edge records 45,000 words exchanged between two collaborators over a quarter.
Community Detection
A class of graph algorithms that identify groups of nodes more densely connected to each other than to the rest of the network, revealing natural clusters within an organization.
Example: Running community detection on a collaboration graph reveals three tightly knit groups that correspond to informal working teams spanning formal department boundaries.
Connected Components
Maximal subgraphs within an undirected graph in which every node is reachable from every other node, with no edges connecting nodes in different components.
Example: An email network with three connected components indicates three groups of employees who communicate internally but have no direct email contact across groups.
Continuous Improvement
An ongoing, systematic process of making incremental enhancements to organizational analytics practices, data quality, models, and workflows based on feedback, measurement, and iterative refinement.
Cosine Similarity
A similarity measure computed as the cosine of the angle between two vectors, yielding a value between -1 and 1, where 1 indicates identical orientation regardless of magnitude.
Example: Two employees whose skill-profile vectors point in nearly the same direction in a high-dimensional space would have a cosine similarity close to 1.
Cross-team Interaction
Communication, collaboration, or knowledge exchange that occurs between members of distinct organizational teams, measured by the frequency and nature of inter-team connections in a network.
Example: Slack messages exchanged between the engineering team and the design team represent cross-team interaction.
Cypher Query Language
A declarative, pattern-matching query language developed by Neo4j for creating, reading, updating, and deleting data in property graph databases using ASCII-art syntax to represent nodes and relationships.
Example: MATCH (e:Employee)-[:REPORTS_TO]->(m:Manager) RETURN e.name, m.name retrieves all employee-manager reporting pairs.
Dashboard Design
The discipline of creating effective visual display layouts that organize charts, metrics, filters, and interactive elements to communicate analytical findings clearly and support decision-making.
Example: A dashboard groups retention metrics in the top row, engagement indicators in the middle, and drill-down filters along the left sidebar.
Data Consent
Explicit, informed permission granted by an individual authorizing the collection, processing, and use of their personal data for specified purposes.
Example: During onboarding, a new employee signs a consent form permitting the organization to include their communication metadata in aggregate network analytics.
Data Encryption
The process of transforming data into an unreadable format using a cryptographic algorithm and key, ensuring that only authorized parties with the correct decryption key can access the original content.
Example: Employee salary properties stored in a graph database are encrypted at rest using AES-256, preventing unauthorized access even if storage media are compromised.
Data Ingestion Pipelines
Automated, multi-stage workflows that extract data from source systems, apply transformations and quality checks, and load the processed records into a target data store in a repeatable sequence.
Example: A pipeline extracts badge-swipe records from the access control API, normalizes timestamps to UTC, validates employee IDs against the master list, and writes the events as nodes in the graph database.
Data Minimization
The principle of limiting data collection and retention to only the information strictly necessary for a stated purpose, reducing privacy risk and storage burden.
Example: A communication analytics system captures message timestamps and participants but deliberately excludes message body content, collecting only what is needed for network analysis.
Data Quality Checks
Systematic validation procedures applied to incoming or stored data to verify its accuracy, completeness, consistency, timeliness, and conformity to defined schema rules before downstream use.
Example: A quality check rejects an employee event record where the timestamp is in the future or the employee ID does not match any known node in the graph.
Data Visualization
The representation of data and analytical results through graphical formats such as charts, graphs, network diagrams, and maps to facilitate comprehension, pattern recognition, and insight communication.
Example: A force-directed graph layout displays departmental collaboration intensity using edge thickness proportional to interaction frequency.
Decision Shapers
Individuals who substantively influence organizational decisions through informal channels such as advice-giving, agenda-setting, or coalition-building, without necessarily holding formal decision-making authority.
Example: A respected senior engineer whose technical opinions consistently sway architectural decisions, despite not being on the architecture review board, is a decision shaper.
Deduplication
The process of identifying and removing or merging redundant records within a dataset so that each real-world entity or event is represented exactly once.
Example: When both the HRIS and the badge system create a node for the same employee, deduplication merges them into a single node using the corporate ID as the matching key.
Degree Centrality
A graph metric that measures a node's importance by counting the total number of edges directly connected to it, normalized by the maximum possible connections.
Example: An employee node with 47 direct communication edges in a 200-person network has a degree centrality of 47/199, indicating broad connectivity.
Department Structure
The formal arrangement of departments within an organization, represented in a graph as hierarchical or lateral relationships among organizational unit nodes.
Example: A graph models the Engineering department node as a child of the Technology division node, with sub-department nodes for Frontend, Backend, and QA beneath it.
Depth-first Search
A graph traversal algorithm that explores as far as possible along each branch before backtracking, following edges from the current node to unvisited neighbors recursively.
Example: DFS starting from a project node follows the chain of task dependencies to their deepest level before backtracking to explore alternative dependency branches.
Desktop Activity
Event data generated by interactions with desktop computing environments, including application launches, file operations, window focus changes, and idle periods.
Example: A desktop activity log records that an analyst opened a spreadsheet at 9:02 AM, switched to a browser at 9:15 AM, and returned to the spreadsheet at 9:31 AM.
Detecting AI Events
The process of identifying which events or content within organizational data streams were generated or significantly influenced by artificial intelligence systems rather than direct human action.
Example: A classifier flags meeting transcripts that were auto-summarized by an AI tool, distinguishing them from human-written notes in the event log.
Device Activity Logs
Chronological records produced by hardware endpoints that capture power state changes, network connections, peripheral usage, and other device-level operational events.
Example: A laptop's device activity log records Wi-Fi connections to the office network at 8:47 AM and disconnection at 5:12 PM each workday.
Dijkstra Algorithm
A classic shortest-path algorithm that finds the minimum-cost route from a source node to all other nodes in a weighted graph with non-negative edge weights.
Example: Dijkstra's algorithm finds the fastest information propagation path from the CEO to a field office by traversing weighted communication edges representing average response times.
Directed Acyclic Graphs
Directed graphs in which no sequence of edges allows traversal from a node back to itself, ensuring a strict ordering without circular dependencies.
Example: An organizational approval chain where requests flow from requester to supervisor to director to VP forms a directed acyclic graph because approvals never cycle back to an earlier stage.
Directed Graphs
Graphs in which each edge has a defined start node and end node, establishing a one-way relationship between the two connected entities.
Example: An edge labeled REPORTS_TO pointing from an Employee node to a Manager node indicates the direction of the reporting relationship.
Disengagement Signals
Observable behavioral indicators within organizational data that suggest declining employee involvement, motivation, or commitment, often preceding voluntary departure or reduced performance.
Example: A decrease in Slack message frequency, fewer code commits, and skipped optional meetings form a cluster of disengagement signals for a software engineer.
Edge Properties
Named attribute-value pairs stored directly on an edge in a property graph, describing characteristics of the relationship that the edge represents.
Example: A COLLABORATED_WITH edge between two Employee nodes carries properties such as project_name: "Q3 Migration" and hours_shared: 120.
Edges
Elements in a graph data model that represent relationships or connections between two nodes, optionally carrying a type label and associated properties.
Example: A MENTORS edge connects a senior engineer node to a junior engineer node, representing their mentoring relationship.
Efficiency Metrics
Quantitative measures that evaluate how effectively an organization converts inputs such as time, effort, and resources into desired outputs, often derived from process timing and communication patterns.
Example: Average time from request submission to resolution, measured across a support network, is an efficiency metric for the help desk process.
Eigenvector Centrality
A graph metric that scores a node's importance based not only on the number of its connections but also on the importance of the nodes it connects to, computed iteratively.
Example: A mid-level manager connected to three executives scores higher in eigenvector centrality than a peer connected to three junior staff, despite having equal degree.
Email Event Streams
Continuous sequences of timestamped records generated by email systems, capturing message sends, receives, opens, replies, forwards, and attachment interactions.
Example: An email event stream records that a manager sent a message to five direct reports at 10:15 AM and received three replies within the hour.
Emotion Detection
The identification of specific discrete emotions such as joy, anger, frustration, or surprise expressed in text or speech, going beyond positive-negative polarity to classify affective states.
Example: Analyzing employee survey responses to detect that 40% express frustration and 25% express optimism about a recent reorganization.
Employee Attributes
The set of properties stored on employee nodes in a graph database, describing characteristics such as name, hire date, department, skills, and employment status.
Example: An employee node carries properties {name: "Priya Sharma", hireDate: "2023-03-15", department: "Analytics", level: "Senior"}.
Employee Data Rights
Legal and regulatory entitlements that protect employees regarding how their personal information is collected, processed, stored, shared, and deleted by their employer.
Example: Under GDPR, an employee has the right to request a copy of all personal data the organization holds about them, including data derived from graph analytics.
Employee Event Streams
Continuous flows of timestamped records that capture the full range of an employee's observable workplace activities across multiple source systems over time.
Example: An employee's event stream for a single day includes a badge swipe at 8:30 AM, a login at 8:35 AM, fourteen emails, three meetings, and a logout at 5:45 PM.
Employee Identifier
A unique, persistent value assigned to each employee that serves as the primary key for the corresponding node in the organizational graph, ensuring unambiguous reference.
Example: Employee node emp-00472 uses a system-generated UUID rather than a Social Security number to avoid storing sensitive identifiers as graph keys.
End-to-end Pipeline
A complete, integrated data processing workflow that spans from initial source data ingestion through transformation, graph loading, analysis, and delivery of actionable insights to stakeholders.
Example: Raw HRIS exports flow through validation, deduplication, graph ingestion, community detection, and finally render as a network health dashboard.
Ethical Frameworks
Structured, principled approaches for evaluating and guiding decisions about the responsible use of data, algorithms, and analytics in organizational contexts.
Example: An organization adopts a framework requiring that every new analytics project pass fairness, accountability, transparency, and privacy reviews before deployment.
Ethics of Privacy
The branch of applied ethics concerned with moral principles governing the collection, use, and disclosure of personal information, balancing organizational insight against individual dignity.
Example: An ethics review board evaluates whether monitoring communication patterns to detect burnout risk respects employee autonomy and proportionality principles.
ETL for Graph Data
The process of extracting data from source systems, transforming it into nodes, edges, and properties conforming to a graph schema, and loading the result into a graph database.
Example: An ETL job reads employee and department tables from an HRIS, transforms each employee row into a node and each department assignment into a BELONGS_TO edge, then loads both into Neo4j.
Event Enrichment
The process of augmenting a raw event record with additional contextual information derived from reference data, lookup tables, or complementary systems.
Example: A raw badge-swipe event containing only an employee ID and timestamp is enriched with the employee's department, office location, and manager name from the HRIS.
Event Logs
Ordered collections of discrete, timestamped records that document the occurrence of specific actions or state changes within a system or process.
Example: An HRIS event log contains sequential entries for an employee's hiring, role change, promotion, and department transfer over a three-year period.
Event Normalization
The process of converting event records from heterogeneous source formats into a single, standardized schema with consistent field names, value types, and units.
Example: Badge events using "emp_num" and email events using "employee_id" are both normalized to a common "employee_identifier" field with a consistent format.
Executive Dashboards
High-level visual summary displays designed for senior leadership that present key organizational performance indicators, trends, and alerts in a consolidated, immediately interpretable format.
Example: The CHRO's dashboard shows headcount trends, attrition risk scores, diversity metrics, and succession readiness on a single screen.
Feature Engineering
The process of selecting, transforming, and constructing informative input variables from raw data to improve the predictive performance of machine learning models.
Example: Deriving a "meeting-to-email ratio" feature from calendar and email data to predict employee collaboration style.
Flight Risk Detection
The use of analytical models applied to employee behavioral data, engagement patterns, and contextual factors to estimate the probability that specific individuals will voluntarily leave the organization.
Example: An employee whose collaboration network has shrunk by 50% and who has not received a promotion in four years is flagged as high flight risk.
Fragmentation Analysis
The measurement and assessment of disconnection or weak connectivity within an organizational network, identifying the degree to which the organization has splintered into poorly communicating segments.
Example: After a merger, fragmentation analysis reveals that legacy Company A employees and legacy Company B employees form two loosely connected clusters with only five bridging relationships.
Graph Algorithms
Computational procedures that operate on graph structures to discover patterns, calculate metrics, find paths, or identify communities within node-and-edge data.
Example: An organizational analyst runs centrality, community detection, and pathfinding algorithms on a collaboration graph to surface hidden influencers and communication bottlenecks.
Graph Classification
A machine learning task in which an entire graph is assigned to a predefined category based on its structural and attribute properties, rather than classifying individual nodes or edges.
Example: Classifying organizational communication graphs as "healthy" or "at-risk" based on their overall topology and interaction patterns.
Graph Data Model
A formal representation of how information is organized in a graph database, defining the types of nodes, edges, properties, and constraints that constitute the schema.
Example: A graph data model for organizational analytics defines Person and Department node types connected by BELONGS_TO and REPORTS_TO edge types, each with specified properties.
Graph Database Performance
The speed and resource efficiency with which a graph database executes queries, traversals, and write operations, influenced by data volume, query complexity, and indexing strategies.
Graph Databases
Database management systems that use graph structures with nodes, edges, and properties to store, query, and manage data, optimized for traversing relationships between entities.
Example: Neo4j stores each employee as a node and each reporting relationship as an edge, enabling queries like "find all people within three hops of the CEO" in milliseconds.
Graph Library Design
The architectural planning and structural organization of reusable graph database components, including query interfaces, algorithm wrappers, and data access patterns optimized for maintainability and extensibility.
Example: The library exposes a standard interface where any centrality algorithm can be called with a subgraph filter and returns a ranked node list.
Graph Machine Learning
A family of machine learning techniques specifically designed to operate on graph-structured data, leveraging node attributes, edge relationships, and topological patterns for prediction and classification tasks.
Graph Metrics
Quantitative measures that describe structural properties of a graph, including characteristics such as density, diameter, clustering coefficient, and centrality distributions.
Example: Reporting that a collaboration network has a density of 0.12, a diameter of 7, and an average clustering coefficient of 0.45 provides a graph metrics summary.
Graph Neural Networks
A class of neural network architectures that operate directly on graph structures by iteratively aggregating and transforming feature information from neighboring nodes to learn node, edge, or graph-level representations.
Example: A graph neural network trained on an organizational network can predict which employees are likely to leave by learning from the structural patterns around departed employees.
Graph Query Language
A category of formal languages designed to express pattern-matching, traversal, and manipulation operations against data stored in graph database structures.
Example: Cypher, Gremlin, and SPARQL are graph query languages, each offering syntax for matching patterns of nodes and edges.
Graph Scalability
The ability of a graph database system to maintain acceptable performance and availability as the number of nodes, edges, concurrent users, or query complexity increases.
Graph Schema Design
The practice of defining the structure of a graph database by specifying node labels, edge types, property keys, data types, constraints, and indexing strategies to support intended queries.
Example: A schema design for HR analytics defines Employee and Role node labels, a HAS_ROLE edge type with a start_date property, and a uniqueness constraint on employee ID.
Graph Traversals
Operations that navigate from a starting node through connected edges and neighboring nodes, following defined patterns to discover paths, subgraphs, or reachable entities.
Example: A three-hop traversal from the CEO node along REPORTS_TO edges returns all employees within three levels of the organizational hierarchy.
Graph vs Relational
A comparative analysis of graph databases and relational databases, evaluating their respective strengths in data modeling, query expressiveness, relationship traversal performance, and schema flexibility.
Hidden Achievements
Employee accomplishments that are not captured by formal recognition systems or standard reporting channels but are discoverable through analysis of collaboration patterns, event streams, or peer interactions.
Example: Graph analysis reveals an engineer who informally mentored twelve new hires, none of whom appeared in the official mentoring program records.
HRIS
A software system that manages core human resources functions including employee records, payroll, benefits administration, time tracking, and regulatory compliance reporting.
Example: Workday and SAP SuccessFactors are HRIS platforms that serve as the system of record for employee demographic and employment data.
Human Resources Data
Structured information about an organization's workforce, encompassing employee demographics, compensation, job history, performance evaluations, and organizational assignments.
Example: An HR dataset includes records showing each employee's hire date, current role, department, salary band, and performance rating.
Idea Flow Networks
Graph structures representing the paths, intermediaries, and channels through which ideas propagate across an organization, from origination through refinement to adoption or rejection.
Example: A network visualization shows that most adopted product ideas pass through two specific cross-functional connectors before reaching the decision committee.
Ideation Tracking
The systematic monitoring and recording of idea generation, submission, evaluation, and progression through organizational workflows, enabling measurement of creative contribution and innovation pipeline health.
Example: A graph tracks each idea node from initial proposal through peer review, pilot approval, and implementation status.
Inclusion Analytics
The measurement and analysis of equity in organizational network access, participation patterns, communication flows, and opportunity distribution across demographic groups and structural positions.
Example: Analysis reveals that remote employees have 35% fewer cross-departmental connections than on-site peers, indicating a network inclusion gap.
Indegree
The count of incoming directed edges terminating at a node, indicating how many other nodes point to or reference that node.
Example: A manager node with an indegree of 8 on REPORTS_TO edges has eight direct subordinates in the organizational hierarchy.
Indexing in Graphs
The creation of auxiliary data structures within a graph database that map property values to nodes or edges, enabling direct lookups without full graph scans.
Example: Creating an index on the employee_id property of Employee nodes allows the database to locate a specific employee in constant time rather than scanning all nodes.
Influence Detection
The process of identifying individuals within an organizational network who disproportionately shape the opinions, behaviors, or decisions of others, using metrics such as centrality, information flow, and network position.
Example: Analyzing email reply patterns to discover that a mid-level manager's messages consistently trigger action across multiple departments indicates high influence.
Informal Leaders
Individuals who exert significant influence over colleagues through expertise, trust, or social capital rather than through formally assigned managerial or supervisory authority.
Example: A software developer whom teammates consistently consult before making technical decisions, despite having no management title, functions as an informal leader.
Information Flow Analysis
The study of how information propagates through an organizational network, tracing paths, speeds, and transformations of messages from origin to destination across nodes and edges.
Example: Mapping how a policy announcement travels from the executive team through middle management to front-line staff, measuring time delays at each layer.
Innovation Metrics
Quantitative and qualitative measures of an organization's creative output, including idea volume, adoption rates, time-to-implementation, novelty scores, and the diversity of contributors in the innovation network.
Example: A quarterly report tracks that 120 ideas were submitted, 18 reached pilot stage, and 5 were deployed to production.
Integration Monitoring
The ongoing tracking and measurement of how teams, processes, and systems combine during organizational changes such as mergers, acquisitions, or departmental consolidations.
Example: Weekly graph snapshots show the growth of cross-team communication edges between two formerly separate engineering divisions after a merger.
Jaccard Similarity
A similarity coefficient defined as the size of the intersection of two sets divided by the size of their union, yielding a value between 0 and 1 that measures the overlap between two nodes' neighbor sets.
Example: If Employee A collaborates with {B, C, D} and Employee E collaborates with {C, D, F}, their Jaccard similarity is 2/4 = 0.5.
Knowledge Concentration
A condition in which critical expertise, institutional memory, or specialized skills are held by an insufficient number of individuals, creating organizational fragility and operational risk.
Example: Only two engineers understand the legacy payroll system's architecture, creating a knowledge concentration risk for the finance department.
Label Propagation
A community detection algorithm in which each node is initially assigned a unique label and then iteratively adopts the most frequent label among its neighbors until labels stabilize, revealing community structure.
Example: Running label propagation on an email network assigns employees to communities based on who they communicate with most frequently, without requiring a predefined number of groups.
Labeling Communities
The process of assigning meaningful, human-readable names or descriptions to groups identified by community detection algorithms, based on shared attributes, functions, or roles of group members.
Example: After detecting a cluster of employees, examining their departments and project assignments to label the community "Cross-functional Product Launch Team."
Large Language Models
Neural network models with billions of parameters trained on extensive text corpora, capable of generating, summarizing, classifying, and reasoning about natural language text across diverse domains.
Example: Using a large language model to automatically draft plain-language summaries of organizational network analysis findings for executive stakeholders.
Latency Management
The set of strategies and techniques used to monitor, minimize, and control the time delay between data generation at the source and its availability for query in the target system.
Example: Switching from nightly batch loading to micro-batch ingestion every five minutes reduces the latency of employee event data from 24 hours to under 10 minutes.
License Tracking
The process of monitoring and recording software license assignments, usage, and compliance as relationships between employee nodes and software asset nodes in a graph.
Example: A HAS_LICENSE edge connects an employee node to a "Tableau Desktop" node with properties for assignment date, expiration, and last-used timestamp.
Link Prediction
A graph-based inference task that estimates the likelihood of a future or missing edge between two nodes based on existing network structure and node attributes.
Example: Predicting that two employees who share many common collaborators but have never directly communicated are likely to form a working relationship.
Login and Logout Events
Authentication records that capture the timestamp, user identity, device, location, and outcome of each session start and session end on an organizational system.
Example: A login event records that employee E-4521 authenticated to the CRM application at 8:42 AM from the Chicago office VPN; the corresponding logout event occurs at 12:15 PM.
Louvain Algorithm
A fast, greedy community detection algorithm that iteratively optimizes modularity by merging nodes into communities and then aggregating communities into super-nodes across multiple passes.
Example: The Louvain algorithm partitions a 5,000-node collaboration graph into 23 communities in under two seconds, revealing informal working groups across the enterprise.
Machine Learning
A branch of artificial intelligence comprising algorithms and statistical models that automatically improve their performance on a task through exposure to data, without being explicitly programmed for each decision.
Meeting Patterns
Recurring structures and trends observed in meeting data, including frequency, duration, participant composition, scheduling habits, and cross-team participation rates.
Example: Analysis reveals that a department averages 22 hours of meetings per person per week, with 40% of meetings having more than eight attendees.
Mentoring Matching
The analytical process of identifying and recommending suitable mentor-mentee pairings based on skill complementarity, career goals, network position, personality compatibility, and organizational context.
Example: A matching algorithm pairs a junior data scientist with a senior analyst who shares her interest in NLP and is located in an adjacent network community.
Mentor-mentee Pairing
The specific operational process of formally assigning a mentor to a mentee, including evaluation of compatibility criteria, goal alignment, availability, and relationship structuring within a mentoring program.
Example: After matching scores are computed, HR pairs 30 new hires with experienced employees, scheduling their first meetings within the onboarding week.
Merger Integration
The comprehensive process of combining two or more organizations into a unified entity, encompassing the alignment of structures, cultures, processes, systems, and personnel networks.
Example: After acquiring a competitor, the company uses graph analytics to identify overlapping roles and bridge employees who connect both legacy networks.
Mobile Device Events
Timestamped records generated by smartphones and tablets that capture application usage, location changes, connectivity transitions, and notification interactions in an organizational context.
Example: A mobile device event log shows a field technician launching the work-order app at a client site, completing an inspection form, and uploading photos over the cellular network.
Modeling Communication
The practice of representing organizational communication as a graph structure, where employee nodes are connected by directed or weighted edges denoting information exchange.
Example: An email communication graph creates a directed EMAILED edge from sender to recipient for each message, with timestamp and thread-id properties.
Modeling Employees
The practice of representing individual employees as nodes in a labeled property graph, with properties capturing personal and professional attributes and edges capturing relationships.
Example: Each employee becomes a :Person node connected to department, role, and project nodes through typed relationships such as BELONGS_TO and ASSIGNED_TO.
Modeling Organizations
The practice of representing an organization's formal and informal structures as a graph, with nodes for units, teams, and divisions connected by hierarchical and lateral relationships.
Example: A company graph contains :Organization, :Division, :Department, and :Team nodes linked by PART_OF edges forming a containment hierarchy.
Modeling Positions
The practice of representing job positions as distinct nodes in a graph, separate from the employees who fill them, enabling tracking of vacancies, succession, and role history.
Example: A :Position node for "Senior Data Engineer" connects to a :Department node via BELONGS_TO and to an :Employee node via FILLS with start and end date properties.
Modeling Projects
The practice of representing projects as nodes in a graph, connected to team members, tasks, milestones, and organizational units through typed relationships.
Example: A :Project node for "CRM Migration" links to employee nodes via WORKS_ON edges and to :Task nodes via CONTAINS edges with status and deadline properties.
Modularity
A scalar measure ranging from -0.5 to 1.0 that quantifies the quality of a network partition into communities by comparing the density of edges within communities to the expected density in a random network with the same degree distribution.
Example: A modularity score of 0.65 for a departmental partition indicates substantially more intra-department communication than would occur by chance.
Motif Detection
The identification and counting of recurring small subgraph patterns, typically involving three to six nodes, that appear in a network more frequently than expected by chance, revealing fundamental structural building blocks.
Example: Detecting a high frequency of feed-forward loop motifs (A to B, B to C, and A to C) in a management communication network suggests systematic information cascading.
Named Entity Recognition
A natural language processing task that identifies and classifies mentions of real-world entities such as people, organizations, locations, dates, and job titles within unstructured text.
Example: Extracting "Jane Park," "Engineering Division," and "Q3 2025" from a project status email as person, organization, and date entities respectively.
Natural Language Processing
A field of artificial intelligence focused on enabling computers to understand, interpret, generate, and respond to human language in both written and spoken forms.
Network Centrality Equity
The principle and practice of ensuring that access to influential, well-connected, or strategically advantageous positions within organizational networks is fairly distributed across demographic groups and roles.
Example: Analysis shows that women hold only 12% of high-betweenness-centrality positions despite comprising 45% of the workforce, prompting targeted networking initiatives.
Network Density
The ratio of the number of actual edges in a graph to the maximum number of possible edges, expressed as a value between 0 and 1, indicating how interconnected the network is.
Example: A team of 10 people with 15 collaboration links has a density of 15/45 = 0.33, meaning one-third of all possible pairings actively collaborate.
Node Embeddings
Fixed-length numerical vector representations of graph nodes that encode structural and attribute information, enabling nodes to be used as inputs to standard machine learning algorithms.
Example: Generating 128-dimensional embeddings for each employee node so that employees with similar network positions cluster together in vector space.
Node Properties
Named attribute-value pairs stored directly on a node in a property graph, describing the characteristics of the entity that the node represents.
Example: An Employee node carries properties such as name: "Priya Sharma", hire_date: "2021-03-15", and department: "Engineering".
Node Similarity
A general measure quantifying how alike two nodes are based on shared properties such as common neighbors, similar attributes, or equivalent structural positions within a graph.
Example: Two project managers with overlapping team connections and similar tenure may have high node similarity even if they work in different divisions.
Nodes
Fundamental elements in a graph data model that represent discrete entities, each identified by a unique internal identifier and optionally carrying labels and properties.
Example: In an organizational graph, individual nodes represent employees, departments, projects, and office locations.
Onboarding Data Model
A graph schema designed to represent the new-hire integration process, capturing relationships among new employees, mentors, training modules, milestones, and required tasks.
Example: A new employee node connects to an onboarding checklist node via ASSIGNED_CHECKLIST, with individual task nodes linked by REQUIRES edges encoding completion order.
Onboarding Effectiveness
A measure of how successfully new employees are integrated into an organization, assessed through metrics such as time-to-productivity, network connectivity growth, early engagement levels, and retention within the first year.
Example: New hires who reach ten distinct collaboration connections within 30 days show 25% higher retention at the one-year mark.
Operational Reports
Regularly produced documents or displays that present standard organizational metrics, activity summaries, and performance indicators to support routine monitoring and management decisions.
Example: A weekly operational report shows headcount changes, open requisitions, training completions, and average time-to-fill across all departments.
Optimal Task Assignment
The process of matching individuals to tasks by evaluating capabilities, availability, workload balance, skill development goals, and task requirements to maximize both performance and employee growth.
Example: A constraint-satisfaction algorithm assigns code review tasks to developers whose expertise matches the codebase area while balancing weekly review loads.
Organization Attributes
The set of properties stored on organizational unit nodes in a graph, describing characteristics such as unit name, type, cost center, location, and headcount.
Example: A department node carries properties {name: "Product Design", costCenter: "CC-4010", location: "Austin", headcount: 34}.
Organizational Analytics
The discipline of applying data analysis, graph modeling, and computational methods to organizational data in order to discover patterns, measure performance, and support evidence-based decisions about workforce and structure.
Example: Analyzing communication graphs and event streams reveals that two departments rarely collaborate despite a shared strategic objective.
Organizational Network Analysis (ONA)
A subset of organizational analytics focused specifically on mapping and measuring the informal networks of communication, collaboration, and influence among people within an organization. ONA uses graph-based methods — centrality, community detection, pathfinding — to reveal hidden connectors, bottlenecks, and silos that formal org charts do not capture. While organizational analytics encompasses the full spectrum of graph modeling, data pipelines, NLP, machine learning, and dashboard design covered in this course, ONA refers to the network-mapping practice at its core.
Example: An ONA study reveals that a mid-level project manager with no direct reports has the highest betweenness centrality in the company, bridging Engineering and Product teams that would otherwise be disconnected.
Organizational Health Score
A composite metric that aggregates multiple indicators of organizational wellbeing, including engagement, collaboration density, knowledge distribution, turnover risk, and alignment, into a single summary measure.
Example: The quarterly health score combines network connectivity (0.82), engagement index (0.71), and knowledge distribution (0.65) into a weighted composite of 0.74.
Organizational Hierarchy
The formal reporting structure of an organization represented as a directed acyclic graph, where edges denote supervisory authority flowing from higher-level to lower-level nodes.
Example: A REPORTS_TO edge from an analyst node to a director node, and from the director node to a VP node, encodes two levels of the management hierarchy.
Outdegree
The count of outgoing directed edges originating from a node, indicating how many other nodes that node points to or references.
Example: An employee node with an outdegree of 15 on EMAILED edges sent messages to 15 distinct recipients during the analysis period.
PageRank
An iterative algorithm that assigns importance scores to nodes based on the quantity and quality of incoming edges, where links from highly ranked nodes contribute more weight.
Example: Running PageRank on an internal knowledge-sharing graph reveals that a technical writer receives links from many authoritative sources, ranking her as the top knowledge hub.
Pathfinding Algorithms
A family of graph algorithms that compute routes between nodes, optimizing for criteria such as fewest hops, lowest cost, or shortest weighted distance.
Example: A pathfinding algorithm traces the shortest communication path from a field engineer to the chief architect, revealing three intermediary handoffs.
Pattern Detection
The process of identifying recurring structures, sequences, or trends within organizational data through statistical methods, graph algorithms, or machine learning techniques.
Example: Temporal analysis detects a recurring pattern where cross-team collaboration drops 30% in the two weeks preceding each quarterly review cycle.
Placement Optimization
The analytical process of determining the best assignment of individuals to roles by evaluating the fit between person attributes and position requirements to maximize organizational performance and individual satisfaction.
Example: A graph model evaluates skill overlap, team network compatibility, and career trajectory alignment to recommend the top three candidates for an open architect role.
Privacy by Design
A systems engineering approach that embeds data protection principles into the architecture and design of information systems from inception, rather than adding them as afterthoughts.
Example: An organizational analytics platform is architected so that employee communication graphs are aggregated at the team level by default, with individual-level views requiring explicit authorization.
Process Conformance
The practice of comparing observed process executions, as recorded in event logs, against a reference process model to identify deviations, violations, and compliance gaps.
Example: Conformance checking reveals that 18% of expense reports skip the required manager-approval step and go directly to the finance team.
Process Discovery
The automated extraction of a process model from event log data, producing a visual or formal representation of how activities actually occur without relying on predefined documentation.
Example: Applying a discovery algorithm to onboarding event logs produces a process map showing that IT provisioning and badge issuance happen in parallel rather than sequentially.
Property Graph Model
A graph data model in which both nodes and edges can carry an arbitrary set of key-value property pairs in addition to their structural role, enabling rich, self-describing data representation.
Example: In a property graph, an Employee node has properties like name and title, while a REPORTS_TO edge connecting it to a Manager node has a property indicating the effective date.
Pseudonymization
A data protection technique that replaces directly identifying attributes with artificial identifiers, allowing re-identification only through a separately maintained mapping key.
Example: Employee names in a communication graph are replaced with tokens like "EMP-A7X3," with the mapping table stored in a separate, access-controlled system.
Real-time Data Ingestion
The continuous loading of data into a target system with sub-second to low-second delay from the moment of generation, enabling near-instantaneous query availability.
Example: Badge-swipe events appear as nodes in the graph database within two seconds of the employee tapping their badge at the door reader.
Real-time Discovery
The process of generating analytical insights from organizational data as it is ingested, enabling immediate identification of emerging patterns, risks, or opportunities without waiting for batch processing cycles.
Example: As a resignation event enters the system, real-time queries instantly assess the departing employee's network centrality and flag affected teams.
Recognition Events
Recorded instances of formal or informal acknowledgment of employee contributions, achievements, or milestones within organizational systems, used as positive signals in engagement and retention analytics.
Example: The event stream captures peer-nominated awards, manager shout-outs in team channels, and annual performance bonuses as distinct recognition event types.
Record Retention
Policies and practices that define how long different categories of organizational data are stored before being archived or permanently deleted, balancing legal, operational, and privacy requirements.
Example: A retention policy specifies that employee communication metadata is kept for three years, after which graph edges older than the threshold are automatically purged.
Relational Database Limits
The inherent constraints of relational database systems when handling highly connected data, including performance degradation from multi-table joins, rigid schemas, and the inability to natively represent variable-depth relationships.
Example: A query to find all employees within five degrees of collaboration in a relational database requires five self-joins on a junction table, resulting in minutes-long execution times.
Relational Databases
Database management systems that organize data into tables of rows and columns, enforce referential integrity through foreign keys, and use SQL as the standard query language.
Example: An Oracle or PostgreSQL database stores employee records in an Employees table with columns for ID, name, department, and hire date.
Reorganization Impact
The measurable effects of structural changes such as team restructuring, reporting line modifications, or departmental realignment on communication patterns, productivity, engagement, and collaboration networks.
Example: After a reorganization, graph analysis shows that average path length between engineering and product teams increased from 1.8 to 3.2 hops.
Reporting
The process of organizing, formatting, and presenting analytical findings from organizational data to stakeholders through structured documents, dashboards, or automated deliverables.
Reporting Lines
Directed relationships in an organizational graph that represent formal manager-to-subordinate authority, typically modeled as REPORTS_TO or MANAGES edges between employee or position nodes.
Example: A REPORTS_TO edge from employee node "Jamal" to employee node "Keiko" indicates that Keiko is Jamal's direct supervisor.
Retention Analytics
The application of data analysis techniques to understand the factors contributing to employee turnover and to develop evidence-based strategies for improving workforce stability and reducing unwanted attrition.
Example: A predictive model combining network isolation scores, compensation data, and tenure identifies the top 50 employees at risk of departure within six months.
Reusable Graph Queries
Pre-built, parameterized graph database queries designed to address common organizational analytics questions, packaged for consistent reuse across projects, reports, and applications.
Example: A parameterized Cypher query accepts a department name and returns the top five employees by betweenness centrality within that subgraph.
Role-based Access Control
A security model that restricts system access based on the roles assigned to users, where each role carries a defined set of permissions governing which data and operations are accessible.
Example: Analysts can query aggregated team-level metrics, managers can view their direct reports' individual data, and only HR administrators can access salary properties on employee nodes.
Roles and Titles
Formal job classifications and designations assigned to positions or employees, often modeled as node properties or as separate nodes linked to position nodes in the organizational graph.
Example: A :Position node has a title property of "Staff Data Scientist" and connects to a :Role node labeled "Individual Contributor - Level 6."
Security
The set of administrative, technical, and physical controls implemented to protect organizational data and graph database systems from unauthorized access, modification, or destruction.
Example: A graph database deployment enforces TLS encryption in transit, AES-256 encryption at rest, network segmentation, and multi-factor authentication for all analyst accounts.
Sentiment Analysis
A natural language processing technique that determines the overall emotional polarity of a piece of text, typically classifying it as positive, negative, or neutral.
Example: Classifying employee feedback comments as positive, negative, or neutral to track morale trends after an organizational restructuring.
Sentiment Scoring
The assignment of a numerical value on a defined scale to represent the sentiment expressed in a piece of text, providing a continuous or ordinal measure of emotional polarity and intensity.
Example: Scoring employee survey responses on a scale from -1.0 (very negative) to +1.0 (very positive), where a response about work-life balance receives a score of -0.3.
Shortest Path
The path between two nodes in a graph that minimizes the total cost, where cost may be measured by hop count, edge weight, or another metric.
Example: The shortest path from "Anika" to "Carlos" in a collaboration graph traverses two intermediate nodes, representing the minimum number of referral introductions needed.
Silo Detection
The identification of organizationally isolated groups within a network that exhibit strong internal connectivity but minimal or no communication with other parts of the organization.
Example: Network analysis reveals that the legal department exchanges fewer than 2% of its communications with any other department, indicating a silo.
Similar Events
Events within an organizational event stream that share comparable patterns in attributes such as type, timing, participants, or outcomes, identified through similarity measures applied to event representations.
Example: Two onboarding sequences that involve the same training modules, similar completion times, and comparable mentor interactions are flagged as similar events.
Similar People
Employees or organizational members identified as having comparable profiles based on attributes such as skills, network position, communication patterns, role history, or behavioral characteristics.
Example: A similarity algorithm identifies three engineers in different offices who share the same certifications, project types, and collaboration patterns.
Similar Roles
Job positions within an organization that exhibit comparable characteristics such as required competencies, reporting structures, interaction patterns, or responsibilities, identified through attribute and network-based similarity measures.
Example: Comparing "Business Analyst" and "Data Analyst" roles reveals 80% overlap in required skills, meeting participation, and cross-functional connections, marking them as similar roles.
Similarity Algorithms
Computational methods that quantify the degree of likeness between two nodes, edges, or subgraphs based on structural properties, shared attributes, or vector representations.
Single Points of Failure
Individuals within an organization who exclusively hold critical knowledge, relationships, or capabilities such that their unavailability would cause significant operational disruption to dependent processes or teams.
Example: A graph query identifies one project manager who is the sole connection between the client team and three internal delivery groups.
Skill Gap Analysis
The systematic comparison of skills required for current and future organizational roles against the skills possessed by the existing workforce, identifying deficiencies that require development or hiring interventions.
Example: Mapping required cloud architecture skills against employee certifications reveals that only 3 of 20 infrastructure engineers hold the needed credentials.
Software Application Logs
Timestamped records generated by enterprise software systems that capture user actions, system events, errors, and state transitions within the application.
Example: A CRM application log records that a sales representative created a new opportunity, updated the deal stage, and attached a proposal document within a single session.
Staging Areas
Temporary storage locations where data is held after extraction from source systems and before transformation and loading into the target database, enabling validation and error handling.
Example: Raw CSV exports from the HRIS are placed in a staging area where they undergo format validation and deduplication before ETL loads them into the graph database.
Strategy Alignment
The degree to which organizational activities, resource allocations, projects, and individual efforts are directed toward and supportive of the organization's declared strategic objectives and priorities.
Example: A graph overlay connecting project nodes to strategic goal nodes shows that two of five major initiatives lack any direct link to current strategic priorities.
Stream Processing
A computational paradigm that ingests, transforms, and analyzes data records individually or in micro-batches as they arrive, rather than accumulating them for periodic batch execution.
Example: A stream processing engine evaluates each incoming badge event in real time, immediately creating a VISITED edge between the employee node and the building-floor node.
Subgraph Analysis
The examination of a selected portion of a larger graph to study localized structural properties, patterns, and relationships without the complexity of the full network.
Example: Extracting and analyzing only the nodes and edges within the marketing department to understand internal collaboration patterns independently of the broader organization.
Succession Planning
The process of identifying and developing internal candidates to fill key leadership and critical roles, ensuring organizational continuity and reducing disruption during personnel transitions.
Example: Graph analysis identifies three managers with strong cross-departmental networks and relevant skill profiles as succession candidates for the VP of Engineering role.
Summarization
The process of condensing a longer text or dataset into a shorter representation that preserves the most important information, key themes, and essential meaning of the original content.
Example: Reducing a 50-page quarterly organizational network report into a two-page executive summary highlighting key findings and recommended actions.
Summarizing Events
The creation of concise, coherent descriptions from sequences of organizational events, distilling patterns, outcomes, and key milestones from detailed event logs.
Example: Condensing an employee's 18-month event history of role changes, training completions, and project assignments into a three-sentence career trajectory summary.
Supervised Learning
A category of machine learning in which models are trained using input-output pairs where the correct output labels are provided, enabling the model to learn a mapping from inputs to known outcomes.
Example: Training a model on historical employee data labeled "stayed" or "departed" to predict future attrition risk for current employees.
Task Assignments
Relationships in a project graph that link employee nodes to task nodes, indicating responsibility for completing specific work items, often with properties for role, effort, and status.
Example: An ASSIGNED_TO edge connects employee "Lin" to task "Migrate User Table" with properties {role: "Lead", estimatedHours: 40, status: "In Progress"}.
Text Classification
A natural language processing task that assigns one or more predefined category labels to a given text document based on its content.
Example: Automatically categorizing employee feedback submissions into topics such as "compensation," "work environment," "management," and "career development."
Tokenization
The process of segmenting text into discrete units called tokens, which may be words, subwords, characters, or sentences, as a preparatory step for further natural language processing.
Example: The sentence "Aria mapped the colony's tunnels" is tokenized into ["Aria", "mapped", "the", "colony", "'s", "tunnels"].
Topic Modeling
An unsupervised natural language processing technique that discovers abstract themes or topics within a collection of documents by identifying co-occurring word patterns across the corpus.
Example: Applying topic modeling to thousands of internal wiki articles reveals recurring themes such as "cloud migration," "customer onboarding," and "compliance procedures."
Training and Evaluation
The paired processes of fitting a machine learning model to a training dataset and then assessing its performance on separate validation or test data to measure generalization ability.
Example: Training an attrition prediction model on three years of historical data and evaluating its accuracy, precision, and recall on the most recent six months of held-out data.
Training Gap Detection
The identification of areas where existing training programs, resources, or participation are insufficient to close known skill deficiencies or to prepare the workforce for emerging organizational needs.
Example: Comparing required data literacy skills against training completion records reveals that 60% of managers have not completed the newly mandated analytics curriculum.
Transparency in Analytics
The practice of openly documenting and communicating the data sources, methods, assumptions, and limitations of analytical processes so that stakeholders can understand and scrutinize results.
Example: A network analysis report includes a methodology appendix explaining which communication channels were included, how edges were weighted, and what algorithmic parameters were chosen.
Trend Analysis
The examination of organizational metrics and patterns over time to identify directional changes, cyclical behaviors, and long-term trajectories that inform forecasting and strategic planning.
Example: A twelve-month trend line shows that average employee network size has steadily declined since the shift to remote work, dropping from 14 to 9 connections.
Turnover Contagion
The phenomenon in which one employee's departure increases the probability of additional departures among connected colleagues, spreading through social and professional network ties within the organization.
Example: After a senior engineer resigned, four teammates who shared strong collaboration ties left within the following three months, consistent with a contagion pattern.
Undirected Graphs
Graphs in which edges have no inherent direction, representing symmetric or mutual relationships where the connection applies equally in both directions between the connected nodes.
Example: A COLLABORATES_WITH edge between two researcher nodes in an undirected graph indicates a mutual working relationship without implying that one initiated it.
Universal Timestamps
Standardized time representations, typically in UTC and conforming to ISO 8601, applied consistently across all event sources to enable accurate temporal ordering and cross-system correlation.
Example: Events from a badge system in New York and a VPN server in London are both recorded in UTC, allowing direct comparison without time-zone conversion errors.
Unsupervised Learning
A category of machine learning in which models identify patterns, groupings, or structure in data without access to predefined output labels, relying solely on the inherent properties of the input data.
Example: Clustering employees into groups based on communication patterns and project participation without pre-specifying what the groups should represent.
Vulnerability Analysis
The systematic identification and assessment of structural weak points within an organizational network, including critical dependencies, fragile communication paths, and concentration risks that could cause operational disruption.
Example: Graph analysis reveals that removing just two nodes disconnects the research division from all other departments, exposing a critical structural vulnerability.
Weighted Edges
Edges in a graph that carry a numerical value representing the magnitude, cost, frequency, or strength of the relationship between two connected nodes.
Example: A COMMUNICATES_WITH edge between two employees carries a weight of 47, representing the number of email exchanges between them during the quarter.
Word Embeddings
Dense, fixed-length numerical vector representations of words in a continuous vector space, where words with similar meanings or usage patterns are mapped to nearby points.
Example: In a trained embedding space, the vectors for "manager" and "supervisor" are close together, reflecting their semantic similarity in organizational contexts.