ENCORE: Equipment Name Coreference Resolution

Language Computer Corporation’s (LCC) Lorify Knowledge Base organizes information from a large quantity of unstructured textual documents to facilitate easy analysis of the information. Usually, entities such as people, organizations, or geo-political entities are described across multiple documents, and when reading about an entity in a single document, the most logical next step an analyst might take is to read the other documents that also mentioning this entity. This grouping of entities is called cross-document entity coreference resolution (CDECR). Each grouping is then inserted into a Knowledge Base, which stores all of the relevant references and aggregated information about the entity. In the Lorify Knowledge Base, each knowledge base entry is a webpage that contains references to all the documents mentioning a single entity. Given a single entity in a document, Lorify links that entity to its corresponding knowledge base entry, which shows the analyst the other relevant mentions of the entity of interest. In this effort, called ENCORE (Equipment Name Coreference Resolution), we propose to focus specifically on coreference for entities related to equipment and technology.