Method And Implementation Of Search Engine Evaluation Based On Artificially Label

Over the past decade, the search engine plays an increasingly important role in people’s daily lives. Meanwhile, it has posed a severe test to the ability of the search engine. Search engine technology has always been a research hotspot of the computer industry. It has attracted continuous research of countless people, though being a relatively narrow field, and the continuous research has a wide range of impact and a long history, which is rarely in the history of computer development.There is no absolute objective standard of the good and bad of the search engine. The starting point and the ending point of the evaluation of the sorting algorithm are both human perception. So, evaluation is not only the basic work of the search engine technology, but also one core work of the search engine technology. Identification is always accompanied by evaluation, the method of evaluating the merits of a search engine can’t be self-assessment of developer or just simply relying on feeling, it should be mutually comparable evaluation. Therefore, evaluating the merits of a search engine is a pressing task and is also an issue of concern of every search company. Evaluating the quality of search engine accurately can make the development of search engine technology more rapidly and improve the algorithm of search engine. The most important is that it can also bring the majority of the people a better search experience, being easier to find the request with less unnecessary troubles.This paper implements evaluating search engine by using the method based on artificially label. It solves the lack of evaluators and small amount of evaluation work by adopting crowd sourcing. It solves evaluation task does not match with the ability of the evaluator by creating user groups to distinguish different users with evaluation capacity. It can improve the accuracy of evaluation and the availability of data by establishing one set of certified system, so that users can improve their evaluation capacity by collecting certifications and obtaining certifications. It solves unable to save site and the complexity of evaluation steps by assembling URL, parsing page, saving pages and so on. Task management is that urgent evaluation task first evaluated and non-urgent task can be suspended at any time. It solves the problem of the lack of flexible scheduling tasks. The task pool can show the highest priority task, so one user will only collect one task at the same time. The task pool will also release long time tasks. It solves the problem of evaluation task distribution and time limit by the task pool. Insert monitor is when evaluators are evaluating, the system will insert some cases which the answers have been known into the task, solving the problem of the lack of task monitor, the high costs of monitor, the difficult of accuracy rate and so on. Automatically adding user can shield topics which having the right answer, it solves the problem of the waste of evaluation manpower, the uncontrollable of the costs and so on. Users can get the data of relevance scoring and the data of contrast scoring by downloading the evaluation data reports. It solves the problem of the data difficult using.Evaluation data can be used to calculate the evaluation indicator, such as DCG, NDCG, ERR, data directly reflects the pros and cons of the search engine’s effect, evaluation data can also be used for machine learning, continuous assessment, sampling research. In order to prove the effectiveness of the system, papers assess the effect and analysis, comparison with the data before the system, proof paper method is feasible and efficient.