About the Module 3 content
This module provides HSPs with AI tools developed within the ALLIES project. It shows details on how these tools work, their features, and how HSPs can effectively integrate them into their systems.
It is split into the following units:
- Unit 1: Multimodal Hashing
- Unit 2: Multimodal Visual Understanding
- Unit 3: ALLIES-MTA
- Unit 4: ALLIES-ASR
- Unit 5: Classification
- Unit 6: Risk-assessment
- Unit 7: Knowledge graph
- Unit 8: Explanation Generator Engine
Unit 1: Multimodal Hashing
In this unit, we generate hash representations for both textual and visual content. The hashing algorithms are designed to yield similar hash outputs for perceptually alike images or texts with typographic similarities.
Find in this unit:
- Examples for generating hash representations.
Unit 2: Multimodal Visual Understanding
The primary goal of the multimodal visual understanding is to provide an accurate and robust tool, that can locate and classify terrorist related content on images and videos..
Find in this unit:
- A solution for locating and classifying images and videos.
Unit 3: ALLIES-MTA
The aim of the multilingual text analysis component is to provide Hosting Service Providers (HSPs) with a powerful tool for processing and analysing texts in multiple languages. This component leverages advanced natural language processing (NLP) techniques to transform raw text data into structured information.
Find in this unit:
- A tool for textual multiple language processing and analysis.
Unit 4: ALLIES-ASR
The multilingual Automatic Speech Recognition (ASR) component aims to provide an efficient and accurate system for transcribing spoken language into written text. This component is capable of automatically detecting the language of the audio input and transcribing it accordingly utilising an advanced model.
Find in this unit:
- A tool that helps detecting language of audio input.
By achieving these objectives, the ASR component enhances the accessibility and usability of spoken content, enabling effective analysis and easier understanding by the HSPs.
Unit 5: Classification
This unit aims to categorise the textual content hosted on the online space regarding its relation to the Terrorism Content Online (TCO) domain
Find in this unit:
- Information on how the module can categorise textual content formats.
Unit 6: Risk-assessment
The risk assessment component aims to assess the level of risk a specific content can be considered Terrorist Content Online (TCO) or not.
Find in this unit:
- Background information on the risk assessment module for HSPs.
Unit 7: Knowledge graph
The knowledge graph tool structures and interlinks crucial information regarding TCO, enhancing the ability to analyse and manage complex data relationships effectively. This tool integrates multifaceted data streams related to terrorism, allowing users to conduct intuitive queries and receive insightful, actionable recommendations in an accessible format.
Find in this unit:
- Explanations and details about knowledge graphs
- A description of the Knowledge Graph ALLIES offers to HSPs in this tool.
Unit 8: Explanation Generator Engine
This tool is designed to enhance the interpretability and transparency of AI-driven analyses in the context of identifying potential terrorist content. By utilising natural language processing (NLP) techniques, the XAI component provides clear, understandable explanations of the AI’s decision-making processes, detailing the identified actors, locations, actions, and modalities associated with the content.
Find in this unit:
- Details and examples about the Explanation Generator Engine.
End of Module 3
Jump to the next module or back!
