Research Of Vertical Search Demand For User Experience Requirement

With the rapid development of Internet, the amount of information is becoming larger and larger. How to get the information conveniently becomes the most important thing. Original method is based on artificial arrangement like Yahoo. In this method, person arranges the Internet information and put the website in a specific class on the basis of existing classes. When user searches a keyword, the search engine returns the result from tree-structure class. The accuracy of this method is very high, because its fully done by artificial work. But artificial maintenance is too expensive, the user requirement is in varied forms, the increment of Internet information is so rapid. So this method can't work well for requirement.Then the search engine appears. The experience of search engine is getting the required webpage by keyword search. The data source of search engine comes from the whole Internet. The spider crawl the webpage by the link relationship. Then extract the content of webpage, make the index by keyword. Then calculate the page rank by iterate algorithm. Then provide a user interface, user can search by providing the keyword. The search engine originates from full-text search theory.Vertical search is the further improvement of page search. It base on specific object. Such as map, audio, image, video and so on. Searching a specific object in the vertical search is much better than searching in the page search. The appearance of vertical search meets some user requirement in a way. But the presupposition is that the user must know much about the attribute of the object. For example, when searching a laptop, the user must search the specific parameters such as hard drive, memory, CPU, etc. The user don't know the parameters of laptop can't do anything.This paper implements an vertical search engine which user can get the result by providing the requirement of experiences without knowing the object attribute.Since it search the requirement of experience. So we must think over how to let the user give their experience requirement. This paper let the user give his experience requirement by natural language and digital price information. Such as:rapid speed, portable, etc. This search method is fully different from the origin method which must provide the specific parameters such as memory, hard drive, etc, and it's the innovation point of this paper.The implement method of this search engine is classified the user experience requirement and every classification don't have any information about the attribute of the object. Since every object's information is just its every attribute's information. So every experience requirement must associate with the attribute of the object. The associated attribute will be considered critically.Base on this experience, this paper crawls all the laptop data pages from a laptop website including price, configuration and image. Then getting the parameters from the web pages. Complete the important parameter. Because it can't miss. Then we give the ranking relationship between different parameters. We must use that rank the laptop. The difference between different comparisons is so big. Some of it can compare only by parameters and some of it must compare by experience. For example, the measurement of memory depending on memory size and type. The larger the hard drive, the better it is. But the weight of laptop is the lighter the better. After knowing how to compare the parameter, we must know the influence of every parameter. Because different parameters give the different influence of one effect. For example, CPU gives more influence on the laptop's speed the memory. After finish this work, we must design the human computer interaction.In this process, we categorize different experience. Then make clear how many attributes every class associated. We will raise the influence of an attribute if user's requirement associates with this attribute. The text classifier use VSM model, and Rocchio algorithm, and cosine similarity. User also have price requirement. So we will choose the laptop which price not 500 more expensive and 500 more cheap, then calculate these laptop's weight and rank them. Finally, return the result to the user.In this way, user can get his result by providing the experience requirement without understanding the attribute of the laptop. It's has been improved greatly compared with the old search engine. This means, vertical search can search the object without depending on the attribute of the object. This idea can be used in every vertical search and can make greatly improved.