In recent years, answering keyword queries on graph-structured data has emerged as an important research topic. There are two main reasons behind that. On one hand, at the beginning of their search process, many users are uncertain what exactly they are looking for, or which terms should be used to express their query. So instead of using a query language with well defined terms, they prefer to use keywords coming from their mind to build the query, which gives a more intuitive way of specifying information needs. As to the retrieval results, most people like a list of matching results, ranked by their possible relevance to the query, instead of a binary result(found/not found). Above situation gives traditional document retrieval methods a good application scenario.
In this research, we want to design a probabilistic relevance framework for keyword retrieval on Linked Open Data. Its evaluated implementation will let users use keywords to build a query on linked data, and return ranked possible results corresponding to the required keywords.
22.01.15 - 10:15