Entity Retrieval through Entity Annotated Corpus
MetadataShow full item record
Entity retrieval is an important part of information retrieval, and a large number of search queries are entity oriented. This thesis focuses on performing entity retrieval using an entity annotated corpus. For this purpose we index the ClueWeb12 corpora, and use the FACC1 to create an index of documents that are linked to entities. We adapt the document centric expert retrieval model (a.k.a. Model 2) to entity retrieval and provide the results for three different test collections. The results show that Model 2 achieves a varying score depending on the test collection. The best scores are achieved for the REWQ ClueWeb dataset, with Model 2 achieving MAP score of 0.2637 and P@10 score of 0.3676. A detailed discussion and analysis of the datasets, and the model are presented in the thesis.