InvestigatorAI: An Open-Source AI-Powered Intelligent Investigation Platform

InvestigatorAI is an open-source AI-powered intelligent investigation platform that integrates facial recognition, voice analysis, named entity extraction, sanctions screening, and knowledge graph technologies to provide investigators with structured evidence analysis and risk assessment capabilities. The project is open-sourced under the AGPL-3.0 license, supporting community collaboration and function expansion, aiming to solve the problems of low efficiency and easy omission of clues in traditional investigation methods.

In the era of information explosion, traditional investigations rely on manual processing of massive data, which is inefficient and prone to missing key clues. InvestigatorAI emerged as an open-source AI-powered platform that helps investigators efficiently analyze evidence, connect clues, and generate structured reports by integrating multiple AI technologies. The project uses the AGPL-3.0 license, allowing free use, modification, and distribution to promote community collaboration and continuous improvement.

InvestigatorAI has multimodal evidence processing capabilities, with core technical modules including:

1. Facial Recognition: Adopts the ArcFace ResNet100 model, improving recognition accuracy through angular margin loss;
2. Voice Analysis: Integrates the ECAPA-TDNN model, supporting voiceprint recognition and voice content analysis;
3. Named Entity Recognition: Uses the DeBERTa-v3 model, extracting more than 21 entity types;
4. Sanctions Screening: Integrates OpenSanctions and global sanctions list databases;
5. Financial Intelligence: Connects to FinCEN and Panama Papers databases;
6. Knowledge Graph: Builds entity relationship graphs via Neo4j, supporting hidden association discovery and reasoning.

The investigation workflow of InvestigatorAI is divided into eight stages:

1. Evidence Input: Accepts multiple forms of evidence such as face images, voice clips, and text;
2. Entity Extraction: Extracts structured information through NER and metadata processing;
3. Evidence Standardization: Unifies the data structure of evidence from different sources;
4. Sanctions Screening: Compares entities with sanctions lists and marks high-risk objects;
5. Relationship Extraction: Analyzes relationships between entities and builds an initial association network;
6. Knowledge Graph Construction: Stores entities and relationships in Neo4j;
7. Risk Analysis: Calculates risk scores and provides priority recommendations;
8. Report Generation: Outputs structured reports containing key findings, risk assessments, and action suggestions.

InvestigatorAI is suitable for multiple scenarios:

• Financial Crime Investigation: Tracks suspicious transactions and identifies money laundering networks;
• Corporate Due Diligence: Identifies potential risks when screening partners;
• News Investigation Reporting: Helps journalists sort out people and relationships in events;
• Legal Litigation Support: Assists lawyers in organizing evidence and evaluating case risks.

The technology stack of InvestigatorAI focuses on performance and scalability:

• Facial recognition uses ArcFace ResNet100 (an industry-leading solution);
• Voice analysis uses ECAPA-TDNN (excellent performance in speaker verification);
• NLP uses DeBERTa-v3 (higher entity recognition accuracy than original BERT);
• Knowledge graph uses Neo4j (native graph database, efficient for complex relationship queries);
• Modular design: Each component can run independently or collaboratively, facilitating expansion and maintenance.

As an AGPL-3.0 open-source project, InvestigatorAI has the following advantages:

• Transparency: Open code for review, building user trust;
• Customizability: Users can modify and expand functions according to needs;
• Collaborative Innovation: Community contributions drive continuous project development;
• Cost-Effectiveness: Reduces technical barriers, allowing more organizations to use advanced investigation tools.

InvestigatorAI integrates multiple AI technologies to provide strong support for investigation work, representing an important direction of AI application in the investigation field. In the future, it is expected to integrate more accurate multimodal fusion analysis, intelligent automatic reasoning capabilities, and user-friendly interactive interfaces, becoming an indispensable assistant for investigators and improving the efficiency and accuracy of investigations.