Zenodo.org will be unavailable for 2 hours on September 29th from 06:00-08:00 UTC. See announcement.

Project deliverable Open Access

Lynx D3.9 Information Retrieval and Recommender Services

Maria Khvalchik; Artem Revenko; Christian Sageder; Víctor Rodríguez-Doncel; Pablo Calleja-Ibáñez

The Lynx project aims at facilitating cross-border compliance for European enterprises. This requires the gathering of relevant regulations, which enables end-users to find answers to their regulatory related needs with ease. In order to provide such accessible means, the Lynx partners brought together their technical expertise in the fields of Information Extraction, Semantic Web, Knowledge Management, and Document Management, to create an integrated solution.

An essential part of the solution provided by Lynx is the automatic analysis of documents in order to facilitate their discovery and retrieval by the end-user. These analyses have three primary objectives:

  •  To extract information contained within the documents relevant to user queries.
  •  To put documents in context, in terms of relations they hold to other documents.
  •  To make documents accessible in different languages.

In order to realize these objectives, Work Package 3 serves to develop a series of services that extract various types of information from the documents and store the information in a well-organized, standards-compliant knowledge graph, which we call the Legal Knowledge Graph (LKG). This LKG will contain not only information about these documents but also information extracted from them, as well as general knowledge from the compliance domain, in the form of controlled vocabularies.

The services that are reported in this deliverable are what we dubbed “Information Retrieval and Recommender Services”, and include four services: Search, Terminology Query, Question Answering and Semantic Similarity. Search service provides a lookup through all the corpora and retrieves related documents corresponding to a given query. Terminology Query enriches the question with linguistic information from domain dependent vocabularies (terminologies) and domain independent vocabularies (dictionaries) to facilitate efficient searching through corpora. Question Answering takes in a natural language question and returns the most promising answer. Finally, the Semantic Similarity service adds extra knowledge that is useful for the applications mentioned above.

All the services have been fully implemented and containerized. Some of these services are the result of ongoing research, and thus this work is already an innovation success.

Files (5.8 MB)
Name Size
Lynx_D3.9 (final).pdf
5.8 MB Download
All versions This version
Views 309309
Downloads 214214
Data volume 1.2 GB1.2 GB
Unique views 289289
Unique downloads 205205


Cite as