Data Analysis Based On Game Gold Trading Market

According to the report of ABI search, an analysis institution in America, the global market size of online game will reach to more than29billion[1]. Game Gold is the central in the online game value chain. In order to collect the real-time information about circulation and exchange of Game Gold, a tool is needed for all entities in the online game industry value chain. A large scale marketof online game has promised an emergence of massive network data. But the researches based on natural language processing technology on this field are still few, including text information presentation technology, synonyms problems processing method, feature terms selecting method, text retrieval technology, text categorization technology, web information extraction technology, and so on.For the above mentioned problem, this paper constructs a virtual professional search engine to get the raw web page set related with online game as the initial research, and then use a text classification method combining the word features of online game to classify the raw web page set and then to get those webpages loaded with trading information, and as a result, to collect data and analyze orders from those web pages, which includes redundancy check and status updates. The main contents of this paper are listed as follow:1. Building a vector space model to process the web pages, and creating methods that combined specific area features to select feature terms and process synonyms words. That could compute and lower the dimension of vector space.2. Constructing a virtual professional search engine based on multiple general search engines to get raw web pages related with online game as the originalobject of study.3. According to K-neighbors text classification, proposing a converted text classification method to classify the raw web page set. This method is based on the analysis of training corpus to compute the similarity in cosine between new texts and training texts. Its implementation is simple and accurate and the cost fo r retraining of training texts is very low. The complexity of calculation in timeand space is within the space of linear changes.4. Using the technology of web extraction based on DOM to get order information is simple and efficient, and makes the collection reliable. Using basic idea of genetic algorithm to inspect the changes of state of the orders that were collected at different times, this technology has the performances of global searchoptimization and efficient parallel computing, and the characteristics of self-organization and adaptive learning. This insures the efficiency and accuracy of the collected orders.5. Building a Game Gold data application platform to provide services suchas supply-and-demand information and real-time information.