The Research Of Evaluation Model Algorithm Of Search Engine Based On Time Series

With the advent of the era of big data, search engine as the primary tool for people to obtain information. The search engine directly determines the quality of the people to obtain the information accuracy, speed and cost. Therefore, the analysis of user satisfaction of the search engine has been an important research direction of information retrieval field. This paper is based on the modern search technology and the correlation between satisfaction evaluation work, through the construction of the entropy model and time series model, analysis the main factors causing satisfaction of search engine.This paper is divided into three parts to study:First, review the most important Webpage search in the evaluation of satisfaction evaluation and Session correlation analysis of two main parts, analyzes the present situation of two kinds of assessment and the existing problems. And the discussion of the information entropy theory and the time series model and the related.In the analysis of the current problems of artificial evaluation based on the satisfaction of needs, analysis of the data source-user behavior log. Fusion, object reconstruction, index selection, dimension flat, extended, multi dimension split operation log, will eventually be stored after data processing in Infobright data warehouse to provide data basis for the following analysis. On this basis, construct the judgment matrix entropy model, in the known conclusions in the case of inverse entropy, through the experiment to determine its feasibility and rationality. In the analysis of location entropy model, designed the time series model is consistent with data characteristics, and construct the satisfaction change the final analysis model, analyzed the contribution of different indicators of satisfaction in Session change in the process of experiment.Finally, the design and development of the analysis system of satisfaction changes based on time series, including the log processing, calculating the entropy contribution, calculation of satisfaction, time series prediction, the storage module 5. Provides suggestions for improvement of the search engine service function and the practical analysis systems. Put forward a set of simple and easy, suitable for cluster computing satisfaction factors analysis algorithm theory.