Data Jacket Retrieval Based on Explicit Semantic Analysis

By using the market of data, we are able to make decision and solve problem for new business based on real-world data to do. When someone needs to find some suitable data employed to ideas that are going to be realized, we can use keywords derived from the ideas as query to search in the market of data. However, sometimes the keywords are not included in the descriptions of data in the database, even for human, it is difficult to choose proper keywords. Therefore, the searching results are usually not satisfying in aspect of either quantity or relevancy. This paper presents two approaches of using Wikipedia-based Explicit Semantic Analysis (ESA) for retrieving relevant data, and compares them with the traditional TF·IDF based full-text retrieval method by experiment.