Research On Automatic Recommendation Method Of Search Engine Evaluation Index

With the development of Internet, people rely on Search Engine more than before. People and the search engine company make much account of the result that the Search Engine applies more and more. As a result, the evaluation of the search results become the focus which each search engine company pays attention to. For the moment, much of the evaluations use the online user behavior data and the indexes made by the company to make experiments. In this way, evaluating the search results. However the efficiency and accuracy of choosing the indexes by hand are insufficient.Because of these reasons, this paper introduces the Recommendation Algorithm to the process of recommending the indexes of evaluating the Search Engines. This paper uses the Collaborative Filtering Recommendation Algorithms to recommend the indexes and combines with the FP-Growth Algorithm and Clustering Algorithm based on the Mutual Information. The algorithm mainly includes the following aspects: evaluation experiment label recommendation, evaluation experiment similarity calculation, evaluation index cluster, evaluation index recommendation. And according to the background, we establish the evaluation model of the recommended results for this paper, and the model includes the recall and precision as well as the correlation analysis of evaluation indexes and evaluation results of the experimental results.We verify the parameters needed by the algorithm and prove the advantages of this algorithm through experiment. On one way, increasing the efficiency and accuracy of choosing the indexes by hand, meanwhile making the experience of choosing the indexes to accumulate by machine learning. On the other way, applying theory basis and experimental method for the problem which is similar to this paper.