Skip to main content

Research Assistant

Overview​

Research Assistant is the platform's AI interaction hub, powered by large language models, LightRAG knowledge graphs, and dual knowledge base integration. It provides AI-assisted services including literature interpretation, research exploration, and model switching, helping users move beyond repetitive tasks like literature organization and framework ideation to focus on core academic research and knowledge discovery.

Eligible Roles​

All logged-in users. Free users have computing power and feature quotas; paid members unlock multi-literature comparison, deep knowledge graph reasoning, multi-model switching, and rich-text output with embedded images.

Core Features​

  • Literature Interpretation: Supports single/multi-paper structured analysis, key point extraction, intelligent Q&A on original texts, and cross-document comparative analysis
  • Research Exploration: Automatically generates research directions, innovative ideas, and complete paper drafts with image-text mixed output; built-in rich-text editor with one-click citation navigation
  • Free Model Switching: Switch AI models across all AI interaction scenarios to match different research content generation needs
  • Conversation History Management: Automatically saves all research conversation drafts; supports creating, searching, and reusing past research plans; supports conversation renaming and deletion for organized research idea management
  • Quick Scene Templates: Built-in standardized research templates across multiple disciplines for one-click launch of various research scenarios
  • Citation Tracing: Bidirectional navigation between in-text citation numbers and reference list entries β€” click to quickly locate the corresponding argument position in the text and trace research evidence
  • One-Click Literature Collection: AI-generated references can be directly added to your personal literature library, stored in a knowledge base or literature category for unified archive management
  • Research Output Archiving: Generated research plans and paper drafts can be saved to various document libraries with one click for categorized management and easy future retrieval and reuse
  • Document Traceability Jump: Analysis result files added from Research Assistant to Cloud Docs support one-click traceability jump back to the original conversation in Cloud Docs, enabling quick review of the AI interpretation context

User Guide​

Access: Click "Research Assistant" in the left menu bar to enter the page directly.

Conversation Management​

Click "New Conversation" in the upper-left corner to start a fresh session; click a past conversation in the left sidebar list to review previous Q&A records.

Hover over a conversation entry and click the "Β·Β·Β·" more actions button on the right to perform the following management operations:

  • Rename Conversation: Select "Rename", enter a custom name and confirm for quick identification of different research topics
  • Delete Conversation: Select "Delete", and after confirmation the conversation will be permanently removed and cannot be recovered

Model Switching​

Use the dropdown in the lower-right corner of the input box to select an AI model. Different models vary in computing power consumption, generation style, and applicable scenarios. After switching, subsequent queries in the current session will use the new model.

Literature Interpretation Mode​

Use Case: You have PDF literature and need to perform tasks such as summarization, innovation point extraction, experiment review, full-text translation, or literature review writing; supports uploading multiple PDFs simultaneously for cross-document comparative analysis to quickly map research threads and differences.

Steps:

  1. Switch to "Literature Interpretation" mode below the input box
  2. Click the upload button and select local PDF files (multi-select supported)
  3. After uploading, preview the uploaded PDF list above the input box, showing file names and page counts; supports deleting individual files or clearing all to re-upload
  4. Enter personalized analysis requirements (optional), such as specifying comparison dimensions like research methods, experimental results, or innovation differences
  5. Select an AI model (optional)
  6. Click Send and wait for the AI to complete full-text analysis and content output

Multi-Document Comparative Interpretation​

Upload multiple PDF papers simultaneously for cross-comparative analysis, suitable for literature reviews, research status overviews, and method comparisons:

  • Batch Upload: After clicking the upload button, hold Ctrl/Command to multi-select PDFs, or drag and drop multiple files into the upload area
  • Comparison Dimensions: Specify comparison directions in the input box, such as research background differences, experimental method comparisons, conclusion similarities/differences, and innovation highlights; if unspecified, the AI automatically generates comparative analysis across core dimensions
  • Result Presentation: Generated content includes individual paper summaries + multi-document comparison tables/analyses, clearly presenting research progress and differences across papers; supports switching between per-document and per-dimension viewing modes
  • Token Note: Multi-document comparison consumes more Tokens based on the total page count of uploaded files; select an appropriate number of documents based on your research needs
  • File Limit: A maximum of 20 PDF files per batch upload / per session; the higher the page count per file and total files, the more Tokens consumed

Citation View​

In the AI-generated interpretation content, citation markers such as [1][2] appear in the text. Each reference in the bibliography includes a "View Citation" entry β€” click to jump to the corresponding citation location in the text for quick evidence tracing.

Research Exploration Mode​

Use Case: No reference literature available; generate research topics, research drafts, paper outlines, experiment designs, and innovative ideas from scratch.

Steps:

  1. Switch to "Research Exploration" mode below the input box
  2. Optionally match the corresponding discipline/academic level, or apply a popular research template from the platform
  3. Enter research keywords, innovation requirements, and constraints
  4. Select an AI model (optional)
  5. Send the command to generate a structured research plan, and continue iterating to refine the content

Add References to Literature Management​

The AI-generated review or research plan automatically lists references at the end. Each reference has a "+Add" button on the right for quick collection to your personal literature library:

  1. Click the "+Add" button next to the target reference
  2. Select a destination from the popup menu:
    • Knowledge Base: Add the reference to the knowledge base for subsequent knowledge retrieval and intelligent Q&A
    • Add to Literature Management: Choose to store in "Root Directory" or "References" category for unified literature resource management

View & Manage Interpretation Results​

Retrieve the generated paper draft from the conversation panel, and browse the full abstract and sections directly on the right. Click the card preview to view the document in full screen, with download and export support. Use the input box at the bottom to ask follow-up questions and iteratively refine article content and research direction.

Add to Document Library​

Structured documents such as complete research plans and review reports can be saved to your personal document library with one click:

  1. Click the "Add to Document Library" icon in the upper-right corner of the generated document, or the document library button at the lower-right of the message bubble
  2. Select a target library from the popup list (e.g., Academic Library, Zhidian Document Library, Wenyuan Document Library, DocHub Huiwen Document Library, Yanzhi Document Library, etc.)
  3. Click "+New" to create a new document library category; after confirmation, the document is saved to the selected location
  4. Document Library Category Management: Users can create unlimited custom document library categories. Created categories support renaming, clearing, and deletion; all documents must be removed from a category before it can be deleted

Document Traceability Jump​

Files added from Research Assistant to the Cloud Docs module support one-click traceability jump back to the original Research Assistant conversation for quick review of the complete AI interpretation context:

  1. Go to the "Cloud Docs" module and find the target document added from Research Assistant
  2. Click the "Document Traceability" button on the document details page
  3. The system automatically jumps to the corresponding historical conversation in Research Assistant and opens the analysis result file, making it easy to review the original Q&A records and complete interpretation process

Exception Scenarios:

  1. If the original Research Assistant conversation has been manually deleted, clicking "Document Traceability" will display a prompt: "Original conversation has been deleted, unable to jump for traceability. You can view the references and generation records retained in the document."
  2. After a document is copied or migrated to another team space, the traceability link is broken and jumping to the original conversation is no longer possible.

View Cited References in Literature Management​

References added to Literature Management can be viewed with detailed citation relationships and literature context in the Literature Management module:

  1. Go to "Literature Management" in the left menu bar and click the target reference
  2. The right-side details page supports switching between tabs such as "Literature Info", "Knowledge Graph", "Cited References", and "Literature Relations" for a comprehensive view of citing papers, references, and related research threads

Feature Notes​

  • All AI operations consume personal Tokens; insufficient computing power can be replenished via point redemption or paid recharge
  • Conversation records are private by default and do not automatically sync to Team Space
  • Content added to Literature Management and Document Library is subject to storage quotas of the respective module's membership plan
  • Multi-document comparative interpretation is a premium feature for paid members; free users can experience single-document basic interpretation
  • Document traceability only supports files added from Research Assistant to Cloud Docs; manually uploaded or otherwise created Cloud Docs do not currently support traceability jumps