Feature List Overview

qKnow adopts a modular design with a clear and highly extensible architecture, encompassing eight core modules: Knowledge Graph, Knowledge Base, Knowledge Q&A, Knowledge Recommendation, Intelligent Writing Assistant, Document Intelligence Review, Smart Chain Workshop, and System Management.

Covering the entire chain of knowledge construction, governance, application, and AI integration, qKnow is ready-to-use and flexible to configure, comprehensively meeting diverse enterprise needs—from basic management to intelligent applications.

No.	Module	Submodule	Function Description	Open-Source Edition	Commercial Edition	Differences in Open-Source Edition
1	System Management	User Management	Supports full lifecycle management of user accounts (creation, editing, deletion, activation/deactivation, search), password reset, and role assignment, enabling unified identity management to meet organizational user governance requirements.	✅ (Included)	✅	/
		Role Management	Provides Role-Based Access Control (RBAC) mechanism, supports custom roles with fine-grained permission configuration, enabling flexible control over functional and data permissions.
		Menu Management	Supports visual configuration of system menus and functional points, multi-level tree structure, and binding with roles for dynamic control of interface and permissions.
		Department Management	Supports hierarchical configuration and maintenance of organizational structures, building an enterprise-level department tree to serve as the foundation for permission allocation and task ownership.
		Position Management	Supports definition of positions and user binding, achieving integrated "person-position-permission" management to enhance alignment between personnel responsibilities and system permissions.
		Dictionary Management	Provides system-level data dictionary management, supporting unified maintenance of public codes such as status, types, and categories, ensuring consistent data standards and standardized front-end display.
		Parameter Settings	Supports centralized configuration and dynamic adjustment of system runtime parameters, enhancing system flexibility and maintainability.
		Notifications & Announcements	Supports publishing, editing, and managing system announcements, operational alerts, and business reminders, enabling targeted push and organization-wide reach of critical information.
		Log Management	Centrally records user operation logs and system runtime logs, supports multi-dimensional search, download, and audit by time, user, operation type, etc., meeting compliance and issue traceability requirements.
2	Knowledge Graph	Knowledge Graph Management	Supports parallel management of multiple enterprise-level knowledge graphs across their full lifecycle (construction, editing, updating, querying). Based on a highly scalable architecture, it enables cross-scenario and cross-system knowledge fusion and unified governance. Through a visual interface and open APIs, it builds a traceable, controllable, and extensible intelligent knowledge hub.	❌ (Not included)	✅	Open-source edition supports only a single knowledge graph.
		Graph Exploration	Provides visual interaction and intelligent analysis capabilities for enterprise knowledge graphs, supporting dynamic browsing, precise search, real-time editing, and result publishing within multi-graph environments—creating a one-stop intelligent hub from "seeing knowledge" to "using knowledge" effectively. 1. Graph Display Supports visual rendering of multi-level, large-scale knowledge graphs, dynamically presenting entity-relation topologies to help users quickly identify knowledge hotspots and association paths. 2. Graph Search Combines keyword search, semantic matching, and graph pattern queries to achieve intelligent, intuitive knowledge retrieval ("what you think is what you get"). 3. Graph Update Supports real-time editing and incremental updates of entities, relationships, and attributes during exploration. Changes take effect immediately, ensuring traceable and auditable knowledge evolution. 4. Graph Publishing Activates verified knowledge graphs into production-ready states via a publishing mechanism. Unpublished graphs cannot be invoked by downstream applications like Q&A or recommendation, ensuring content accuracy and security, enabling controlled transition from "development" to "service" state.	✅	✅	/
		Knowledge Center	Meets enterprise knowledge governance needs by centralizing and intelligently organizing unstructured documents, building a standardized, reusable knowledge input hub. 1. File Categorization Builds multi-level classification systems to aggregate unstructured content, improving efficiency in structuring knowledge and accuracy in downstream applications. 2. File Management Supports unified management of mainstream document and image formats including Word, PDF, TXT, Excel, DWG, PNG, JPG. Provides full-dimensional operations such as online preview, download, view, edit, and delete. Compatible with engineering drawings, technical documents, and other business scenarios, enabling centralized, visual, and governable knowledge asset management.	✅	✅	/
		Knowledge Extraction	Automatically extracts structured knowledge from heterogeneous sources, bridging the gap between unstructured text and structured systems. By flexibly defining knowledge expression patterns, it enables accurate identification and structured output of triples (subject-predicate-object), building enterprise-specific knowledge networks and providing high-quality knowledge supply for upper-layer applications like intelligent Q&A and graph construction. 1. Concept Configuration Supports custom business entity types and attribute systems, clarifying the "subject" and "object" scope in knowledge expressions. Semantic features and recognition rules can be defined, establishing a unified semantic baseline for triple extraction to ensure consistency and scalability in knowledge modeling. 2. Relation Configuration Supports defining "relation" types in knowledge triples and their semantic constraints, including direction, participating entities, and contextual conditions, enabling structured modeling of complex business logic to improve accuracy and explainability of knowledge associations. 3. Unstructured Extraction Leverages large model technology to automatically identify entities and relations from documents, texts, and scanned files, accurately generating structured triple knowledge. Supports multiple formats including PDF, Word, TXT, enabling efficient "text-to-knowledge" transformation. 4. Structured Extraction Supports connecting to structured databases. Through field mapping and rule configuration, automatically converts structured data into standard triple format, enabling seamless migration and integration of system data into knowledge graphs.	✅	✅	Open-source edition uses DeepKE open-source tool.
		Knowledge Fusion	Leverages LLM semantic understanding to intelligently identify duplicate or similar entities and relations from multiple sources, accurately flagging potential redundancies and assisting humans in efficient judgment and confirmation. Implements a collaborative governance model of "AI detection + human oversight" to ensure knowledge graph accuracy and conciseness. 1. Duplicate Knowledge Identification Uses LLM semantic understanding to automatically scan existing knowledge graphs, identifying entities and relations with similar names, attributes, or overlapping connections, generating high-confidence lists of suspected duplicates to improve redundancy detection efficiency. 2. Manual Review & Confirmation Provides a visual comparison interface to display side-by-side details of suspected duplicates (names, attributes, relations, sources), allowing business personnel or knowledge administrators to make informed judgments and confirmations. 3. Duplicate Item Merging Supports one-click merging of confirmed duplicate entities or relations, automatically preserving complete attribute information and updating associated relationships to eliminate redundancy and maintain a clean graph structure. 4. Non-Duplicate Retention For items manually judged as non-duplicates (e.g., homonyms with different meanings, synonyms with different forms), the system automatically retains them and marks the review outcome, preventing accidental deletion and safeguarding knowledge completeness. 5. Operation Audit & Traceability Logs all operations—identification, review, merging—with details including operator, timestamp, and before/after comparisons, enabling auditable, traceable, and controllable knowledge fusion processes.	❌	✅	/
		Knowledge Reasoning	Combines LLM semantic understanding with graph structure analysis to proactively mine unknown entities and implicit relationships within the knowledge graph, uncovering business blind spots and association patterns. Through an "AI suggestion + human verification" mechanism, it continuously expands knowledge boundaries, enabling intelligent evolution and value extension of the knowledge graph. 1. Implicit Relationship Discovery Analyzes existing entity association paths using deep contextual understanding from LLMs, intelligently inferring potential relationships (e.g., "indirect collaboration," "technology derivation") and generating high-confidence reasoning suggestions. 2. Unknown Entity Prediction In cases of broken knowledge chains or incomplete information, the model predicts potentially missing key entities (e.g., unrecorded suppliers, technical terms, events), helping fill gaps in the knowledge network. 3. Human Review & Confirmation All reasoning results require review by domain experts or knowledge administrators via a visual interface, supporting detailed views of context, evidence, and source logic, ensuring new knowledge is accurate and trustworthy before adoption. 4. Automatic Knowledge Completion Supports one-click import of approved reasoning results into the graph, automatically creating entities or relations and tagging the source as "AI Inference," enabling continuous dynamic updates to knowledge assets.	❌	✅	/
		Data Source Management	Supports seamless integration with various structured data systems, providing unified, secure, and configurable data connection capabilities. Offers stable, efficient data input channels for knowledge extraction and graph construction, enabling centralized governance and value realization of heterogeneous enterprise databases. 1. Multi-Source Adaptation Supports flexible integration with mainstream relational databases including MySQL, Oracle, DM8 (Dameng), Kingbase (Renmin Jinshan), meeting data integration needs under domesticated and multi-technology-stack environments. 2. Data Source Configuration Provides a visual connection configuration interface supporting selection of database type, IP/port, account/password, connection pool settings, enabling data source registration and connectivity testing without coding. 3. Connection Management Supports start/stop, edit, test, and delete operations on configured data sources, with real-time monitoring of connection status to ensure stable and reliable data channels. 4. Permissions & Security Supports role-based access control for data sources, encrypts stored connection information, and supports SSL/TLS secure connections to ensure secure and compliant data transmission and access. 5. Integration with Extraction Configured data sources can be directly used in "Structured Knowledge Extraction" workflows, enabling field-to-entity mapping and automated extraction, establishing a direct link from database to knowledge graph.	✅	✅	Open-source edition supports: MySQL, Oracle.
3	Knowledge Base	Knowledge Base Management	Supports parallel management of multiple enterprise-level knowledge bases across their full lifecycle (construction, editing, updating, querying). Through visual interfaces and open APIs, it builds a traceable, controllable, and extensible intelligent knowledge hub.	❌	✅	/
		Knowledge Base Settings	Builds an efficient, accurate, and scalable intelligent knowledge foundation, supporting parameter configuration, permission management, and retrieval optimization, while integrating custom Embedding models and hybrid retrieval strategies. 1. Permission Control Supports configuring knowledge base access permissions by user, role, or department, enabling fine-grained control over "who can see, who can edit" to ensure secure and compliant use of sensitive knowledge. 2. Indexing Mode Offers two indexing modes: High-Quality Mode (high recall, high precision, suitable for core business) and Economy Mode (low resource usage, suitable for massive cold data), balancing performance and cost as needed. 3. Embedding Model Supports custom selection or integration of private Embedding models, adapting to industry terminology and enterprise-specific expressions to improve accuracy and professionalism in vectorized semantic understanding. 4. Retrieval Settings Supports three retrieval modes: vector retrieval, full-text retrieval, and hybrid retrieval. Can be flexibly configured per scenario, defaulting to a semantic + keyword fusion strategy to enhance recall rate and relevance for complex queries.	❌	✅	/
		File Categorization	Builds multi-level classification systems to aggregate unstructured content, improving efficiency in structuring knowledge and accuracy in downstream applications.	❌	✅	/
		File Management	Supports unified management of mainstream document and image formats including Word, PDF, TXT, Excel, DWG, PNG, JPG. Provides full-dimensional operations such as online preview, download, view, edit, and delete. Compatible with engineering drawings, technical documents, and other business scenarios, enabling centralized, visual, and governable knowledge asset management.	❌	✅	/
		File Parsing	Performs deep structural parsing of various documents in the knowledge base, supporting custom text processing and intelligent segmentation strategies to ensure original files are accurately transformed into high-quality semantic units, providing high-fidelity input for vector indexing and knowledge retrieval. 1. Text Preprocessing Rules Supports removing redundant characters, headers/footers, special symbols, and other noise, automatically cleaning text content to improve accuracy in subsequent semantic understanding and retrieval. 2. Custom Segmentation Rules Supports flexible configuration of segmentation logic via delimiters (e.g., headings, line breaks), maximum length, and overlap length to avoid semantic fragmentation and preserve context integrity. 3. Segment Type (Text/QA) For continuous text, automatically segments into reasonably sized knowledge chunks based on semantic coherence, suitable for technical documents, reports, etc. For Q&A or FAQ documents, intelligently identifies question-answer boundaries to generate independent QA knowledge units directly serving intelligent Q&A applications. 4. Multi-Format Compatibility Seamlessly integrates with PDF, Word, TXT, PPT, and other formats, combining OCR technology for scanned documents to efficiently convert unstructured content into analyzable text.	❌	✅	/
		Recall Testing	Helps users accurately evaluate the retrieval effectiveness of knowledge bases by simulating real query scenarios to verify paragraph recall quality. Supports dynamic tuning of retrieval parameters to continuously improve retrieval accuracy and recall rate, ensuring knowledge applications "find everything, find the right things." 1. Paragraph Recall Tests whether the system correctly retrieves relevant text paragraphs from the knowledge base based on actual queries, visually displaying hit results and context to assess semantic understanding and retrieval matching capability. 2. Dynamic Adjustment of Retrieval Parameters Supports real-time adjustment of key parameters such as similarity threshold, top-K return count, and retrieval mode (vector/full-text/hybrid) during testing, enabling quick comparison of recall performance under different configurations for optimal strategy setup. 3. Recall Records Automatically saves query statements, retrieval parameters, hit paragraphs, timestamps, and operator information for each test, forming traceable optimization logs for retrospective analysis and team collaboration in tuning.	❌	✅	/
4	Knowledge Q&A	AI Chat	Deeply integrates LLMs with enterprise multi-source knowledge assets, supporting pre-chat selection of one or more knowledge bases and knowledge graphs as answer sources, enabling "on-demand invocation, precise responses." Whether departmental document libraries or cross-system professional knowledge graphs, they can be flexibly combined to ensure every Q&A is based on the most relevant and authoritative knowledge context. 1. AI Chat Management Supports creation, viewing, renaming, deletion, and pinning of chats, allowing users to freely organize personal chat lists and pin important conversations for improved efficiency and user experience. 2. Multi-Knowledge Source Selection When initiating a chat, users can select one or more knowledge bases or knowledge graphs as the knowledge context for this session, supporting federated retrieval across repositories and graphs to ensure answers cover multidimensional information. 3. Multi-turn Dialogue Suggestions Based on context understanding, automatically recommends follow-up questions, guiding users to explore knowledge deeply, enhancing interaction fluency and information acquisition efficiency. 4. Related Resource Recommendations Intelligently associates and recommends system-related resource documents within answer results, supporting content preview, enabling users to quickly access original materials and enhancing answer credibility. 5. Citation in Responses Labels knowledge sources, attaching original citations to key information. Clicking allows navigation to the original paragraph, ensuring answers are traceable and verifiable, eliminating "black-box outputs."	❌	✅	/
		Resource Configuration	Supports unified management of various unstructured resources, providing high-quality input for "related resource recommendations" in knowledge Q&A. Through structured descriptions and semantic indexing, LLMs can precisely match user queries with associated resources, enabling intelligent association and efficient reuse of drawings, images, documents, and other assets. 1. Multi-Type File Support Supports upload and management of various formats including DWG, JPG, PNG, PDF, Word, Excel, PPT, TXT, covering core enterprise resources such as engineering drawings, images, and documents. 2. Resource Description Entry Supports adding structured description information (title, abstract, etc.) to each resource, assisting LLMs in understanding resource semantics and improving matching accuracy between questions and resources. 3. Semantic Index Construction The system automatically extracts resource content and descriptions, generates vectorized representations using custom Embedding models, and incorporates them into a unified retrieval system, enabling intelligent recall in Q&A scenarios. 4. Integration with Q&A for Recommendations During knowledge Q&A, the LLM automatically matches the most relevant configured resources based on the user's query semantics and recommends them in the response, supporting click-to-view and preview to enhance answer practicality.	❌	✅	/
5	Knowledge Recommendation	Knowledge Recommendation	Leveraging large model technology, supports natural language conversation to intelligently identify user intent and proactively recommend relevant popular, followed, and latest knowledge. Eliminates the need for keyword search, enabling a "speak-and-get" intelligent recommendation experience that allows knowledge to find people, improving information delivery efficiency. 1. Popular Knowledge Recommendation Automatically identifies high-value content based on metrics such as view count, likes, and shares. Combines user permissions and business scenarios to intelligently push the most popular knowledge entries during conversations. 2. Followed Knowledge Recommendation Tracks updates in real-time for topics, projects, tags, or knowledge entities a user subscribes to. Precisely presents the latest updates and related content when users ask questions or during system push notifications. 3. Latest Knowledge Recommendation Automatically discovers newly published or updated knowledge across knowledge bases. Supports quick access via natural language queries to ensure critical information is never missed. 4. Conversational Knowledge Access Supports colloquial and unstructured queries. The large model automatically understands user intent, matching and recommending the most relevant popular, followed, or latest knowledge without requiring keywords—lowering the usage barrier. 5. Personalized Recommendation Engine Builds user profiles based on roles, historical behavior, and areas of interest to enable personalized ranking and context-aware presentation during recommendations, improving accuracy and practicality.	❌	✅	/
		Knowledge Search	Supports one-stop search across multiple knowledge bases and knowledge graphs. Users can quickly locate relevant information scattered across document libraries, FAQs, project repositories, and multiple knowledge graphs using keywords or natural language queries. Based on unified semantic indexing, it enables fused retrieval of structured and unstructured data, breaking down information silos and improving the comprehensiveness and accuracy of knowledge discovery. 1. Federated Retrieval Across Multiple Knowledge Bases Supports simultaneous search across multiple knowledge bases (e.g., document libraries, Q&A repositories, project data), returning integrated results to avoid information fragmentation and improve search efficiency. 2. Joint Query Across Multiple Knowledge Graphs Can connect and jointly query multiple knowledge graphs, enabling deep semantic search based on entities, attributes, and relationships to precisely locate domain-specific knowledge and implicit associations. 3. Full-Text Keyword Search Supports high-precision full-text matching on documents and text, quickly locating paragraphs and files containing keywords. Includes enhanced capabilities like fuzzy matching and synonym expansion. 4. Natural Language Query Supports natural language questions. The large model automatically parses intent and converts it into retrieval commands, enhancing search intelligence. 5. Unified Result Ranking Computes semantic relevance and fuses weights for results from different sources, using a rerank model for intelligent sorting—prioritizing the most relevant and authoritative content.	❌	✅	/
6	Intelligent Writing Assistant	Intelligent Writing	Integrates large model generation capabilities with enterprise knowledge assets to support end-to-end assistance from inspiration to final output. Offers capabilities in general writing, template customization, outline generation, intelligent refinement, and multi-format export—helping users rapidly produce high-quality, compliant, and scenario-specific content. 1. General Writing Supports automatic generation of well-structured, fluent drafts based on natural language instructions, covering common scenarios like summaries, reports, and emails. 2. Custom Template Writing Supports creating and reusing enterprise-specific writing templates to ensure consistent formatting, compliance with standards, and improved approval rates. 3. Outline & Article Generation Can first generate a logically clear writing outline, then expand it into a complete article with one click. Supports user-defined structural adjustments, enabling an efficient "plan first, write later" workflow. 4. Collection Allows users to save high-quality generated content to a personal or team collection for quick reuse, building a personalized writing knowledge base. 5. History Records Automatically saves all writing history, supporting retrieval by time, title, or content keywords—ensuring traceability and preventing content loss. 6. Continuation, Expansion, Summarization, Refinement Offers multi-dimensional text optimization: continue writing from context, expand details, condense content, or refine language (improving professionalism, conciseness, or readability) to meet diverse editing needs. 7. File Export Supports one-click export of generated content into common formats such as Word, PDF, and TXT, preserving formatting for easy reporting, archiving, and sharing.	❌	✅	/
		Template Management	Supports centralized configuration of enterprise-grade writing templates and dynamic data integration. By uploading standard document templates, binding business system APIs, and configuring intelligent prompts, it enables automated content population and personalized generation—achieving a true "configure once, reuse multiple times, data-driven" intelligent writing loop. 1. Template Upload Supports uploading standard documents in Word, PDF, and other formats as writing templates (e.g., project proposals, bid documents, weekly reports), preserving original formatting and structure as the foundation for intelligent generation. 2. API Management Allows binding APIs from internal business systems (e.g., ERP, CRM, project management systems) to automatically retrieve dynamic data (e.g., customer info, project progress, budget) during generation, ensuring real-time accuracy. 3. Prompt Management Supports configuring dedicated prompts (Prompt) for each template to clearly define how data from APIs maps to placeholders in the template. Users can manually adjust prompt logic to achieve intelligent data organization and natural language expression.	❌	✅	/
7	Document Intelligence Review (Compliance Check)	Intelligent Review	Combines large model semantic understanding with rule engines to perform comprehensive intelligent validation of document content. Accurately identifies grammar errors, redundant expressions, and logical inconsistencies, providing issue localization, revision suggestions, and original text comparison—helping users quickly improve text quality and professionalism. 1. Result Presentation Clearly displays the type of issue (e.g., grammar error, inappropriate wording, verbose sentence), corresponding revision suggestions, and original text snippets in a three-column layout for easy understanding and decision-making. 2. Quick Error Location Highlights problematic sections in the document preview. Supports clicking on an issue to automatically jump to the corresponding paragraph—enabling "see it, fix it" editing and significantly improving revision efficiency. 3. Multidimensional Inspection Capabilities Supports grammar correction, terminology standardization, style consistency (e.g., formal vs. informal), sensitive word detection, and readability optimization—meeting diverse needs for reports, technical documents, marketing copy, etc. 4. Customizable Inspection Rules Supports configuration of enterprise-specific terminology libraries, banned word lists, and style guides to ensure checks align with organizational policies and industry standards, enabling personalized quality control.	❌	✅	/
		Rule Library	Supports centralized management of text validation rules for grammar, terminology, style, and compliance. Through flexible configuration and continuous iteration, ensures text checking capabilities align with organizational norms, industry standards, and business scenarios—achieving standardized and automated content quality control. 1. Built-in Inspection Rules Provides rich default rules covering basic grammar, punctuation, common typos, and redundant sentences—ready-to-use to ensure baseline text quality. 2. Custom Rule Configuration Allows users to create and manage custom rules such as enterprise-specific terms, banned words, recommended expressions, sensitive terms, and format standards—ensuring documents meet organizational style and compliance requirements. 3. Rule Categorization & Grading Supports classifying rules (e.g., "Grammar", "Terminology", "Compliance") and setting severity levels (e.g., "Suggestion", "Warning", "Error") for differentiated result presentation. 4. Rule Enable/Disable Allows flexible enabling or disabling of specific rules by scenario, department, or document type. Supports template-level rule binding for fine-grained, "one-document-one-policy" inspection strategies. 5. Integration with Text Inspection All rules are applied in real-time during text checks. Results automatically link to corresponding rule entries, providing traceable and explainable revision rationale to enhance user trust.	❌	✅	/
8	Smart Chain Workshop	Intelligent Agent Applications	Designed to efficiently connect external AI agents with the internal knowledge ecosystem. It allows users to seamlessly integrate intelligent workflows built on low-code AI platforms like Dify into qKnow—enabling a new model of "build once, use across platforms, dynamically collaborate." 1. Agent Application Integration Supports seamless integration of custom workflows (Workflows) created on platforms like Dify via API or plugin, enabling capability reuse without redevelopment. 2. Dynamic Input Parameter Adjustment During agent invocation, supports dynamic configuration and visual editing of input parameters (e.g., variable substitution, context adjustment, knowledge source scope)—flexibly adapting to different business scenarios.	❌	✅	/
		Agent Configuration	Centrally manages and securely stores API-Keys required for integration with the Dify platform. Supports hierarchical permission control, enabling/disabling, and updating of keys—ensuring secure and controllable agent invocation and unified governance of cross-platform capability integration.	❌	✅	/
9	Others	Online Documentation	Provides comprehensive official documentation including deployment, operations, APIs, and best practices—updated regularly with a complete structure.	✅	✅	/
		Technical Support	Offers enterprise-grade technical support services, including dedicated technical contacts, SLA-backed support, and 7x24 or 5x8 availability.	🟡 (Partially included)	✅	Open-source edition relies on community support via Issues.
		Source Code Updates	Provides stable version update channels with upgrade guides and patch notes, ensuring long-term maintenance of compatibility and security.	✅	✅	/