Intelligent Text Fact-Checking Assistant

Workflow Name

Intelligent Text Fact-Checking Assistant

Key Features and Highlights

This workflow, built on the n8n automation platform and integrated with Ollama’s local language models, performs fine-grained sentence segmentation on input text and conducts fact-checking on each individual statement. It ultimately generates a structured summary that includes a list of erroneous statements and an overall accuracy assessment. A key highlight is the incorporation of a bespoke AI model (bespoke-minicheck), enabling efficient and professional automated fact verification with support for multi-round data processing and filtering.

Core Problem Addressed

Verifying the accuracy of factual information in textual content—especially news reports and scientific articles—is often time-consuming and labor-intensive when done manually, and prone to subjective bias. This workflow automates sentence splitting and leverages a specially trained language model to validate facts, significantly improving the speed and accuracy of fact verification. It assists content creators and editorial teams in quickly identifying and correcting misinformation.

Application Scenarios

• Fact-checking manuscripts for news organizations and editorial teams
• Verifying facts in academic publications and research reports
• Ensuring authenticity of user-generated content on content platforms
• Validating accuracy of market research and data analysis reports
• Quality control of textual materials in educational settings

Main Process Steps

1. Input Trigger: Manually triggered or invoked by other workflows, accepting the text to be checked along with reference factual data.
2. Text Segmentation: Using a custom JavaScript code node, the input text is split into individual sentences based on punctuation marks such as periods and question marks, with special handling for dates and list formats.
3. Data Merging: The segmented sentences are combined with the factual reference data to form a list of statements pending verification.
4. Sentence-by-Sentence Fact-Checking: The bespoke-minicheck model from Ollama analyzes each sentence against the facts, labeling them as “true” or “false.”
5. Filtering Erroneous Information: Sentences marked as “false” are extracted, while irrelevant or conversational sentences are ignored.
6. Summary Analysis: The number and content of erroneous sentences are aggregated, and a large language model is invoked to summarize and evaluate the fact-checking results, producing a structured error report.
7. Result Output: A report containing the list of incorrect statements and an overall accuracy evaluation is generated to facilitate rapid identification and correction of factual issues in the text.

Involved Systems and Services

• n8n Automation Platform: Workflow orchestration and node management
• Ollama Local Language Models: Including the bespoke-minicheck custom model and the qwen2.5 large language model
• JavaScript Code Node: Implements the text segmentation logic
• Manual Trigger and Workflow Trigger Nodes: Provide flexible initiation methods for standalone testing and process integration

Target Users and Value Proposition

• Editors and Content Reviewers: Enhance fact-checking efficiency and reduce manual workload
• Content Creators and Journalists: Quickly detect factual errors in drafts to improve reporting quality
• Academic and Research Institutions: Assist in verifying literature and reports to ensure research rigor
• Platform Operators: Strengthen authenticity management of user content and maintain a healthy content ecosystem
• Technical Developers: Can integrate this fact-checking module into larger-scale automated workflows

The Intelligent Text Fact-Checking Assistant deeply integrates AI language models with automated processes, enabling users to perform fast and accurate fact verification of texts, thereby significantly advancing modern content quality control.