Project Deliverables and Summaries
This page collects summaries of historical public deliverables related to the technical themes covered on this website, including multimodal AI, speech and language systems, computer vision, semantic technologies, summarisation, retrieval, and intelligent information systems.
Some of the documents listed here were originally published as public project deliverables. On this site, they are presented through plain-English summary pages designed to make their content easier to browse, understand, and connect with broader topics and use cases.
This section focuses specifically on historical deliverables. For a wider archive of research summaries and paper commentaries, visit the Publications page.
What You Will Find Here
Each deliverable summary is intended to give a clear overview of a technical report without requiring the reader to go through the full original PDF first. Depending on the document, summary pages may cover:
- the main purpose of the deliverable
- the system, method, or infrastructure described
- key components or findings
- why the report matters
- links to related topics and use cases
Available Deliverable Summaries
- MULTISENSOR Final System Evaluation Report
Summary of the final evaluation of the integrated system across journalism, commercial media monitoring, and SME internationalisation use cases. - Basic Techniques for Speech Recognition, Text Analysis and Concept Detection
Summary of an early technical deliverable covering named entity recognition, concept extraction, relation extraction, speech recognition, multimedia concept detection, and machine translation. - Graphic Interfaces and Operational Prototype
Summary of the first operational prototype of the platform, including architecture, repositories, crawler, extraction pipeline, and early demonstrator applications. - Summarisation Infrastructure and Baselines
Summary of the first summarisation infrastructure, including extractive summarisation baselines and datasets prepared for future abstractive methods.
Why These Deliverables Matter
Historical deliverables can still be useful because they show how complex AI-related systems were designed, implemented, and evaluated in practice. They often contain details that do not appear in short papers or general overviews, including architecture decisions, module descriptions, integration logic, evaluation methodology, and infrastructure planning.
For readers interested in how research ideas become working systems, these reports provide a more practical and process-oriented view than a standard academic article.