You are here
Knowledge-based Information Retrieval
Given a document collection, Information Retrieval is the task of returning the most relevant documents for a specified user query.
Approaches in this field aim to improve information retrieval performances by exploiting knowledge extraction techniques.
Indeed, traditional IR approaches solve this task by computing the similarity between terms or possible term-based expansions (e.g., synonyms, related terms) of the query and the documents.
These approaches tend to suffer of known limitations, that we exemplify with the query "astronomers influenced by Gauss": relevant documents may not necessarily contain all the query terms (e.g., the term "influenced" or "astronomers" may not be used at all in a relevant document); similarly, some relevant documents may be ranked lower than other ones containing all three terms, but in an unrelated way (e.g., a document about some astronomer, containing the information that he was born centuries before Gauss and was influenced by Leonardo Da Vinci).
This challenge is faced by expanding queries and documents with semantic terms obtained by processing them with Natural Language Processing methods (e.g., Entity Linking, Frame Detection), and by linking them to available Semantic Web and Linked Open Data knowledge bases (e.g., DBpedia, YAGO).