Research And Realization About Service Platform Of Personalized Hydrological Information Aggregation Based On Web3.0

Web3.0 is a new model that was proposed during the development of Internet.The purpose of Web3.0 is to resolve the problem that users can not quickly obtain useful information from massive and dispersed information. Based on Web2.0, Web3.0collects various information scattered all over the Internet, and can meet demands of different users. In the era of Web3.0, the services that users get are more personalized,accurate and intelligent. Based on the study of Web3.0, this paper focused on the realization of information aggregation and personalized service, and its application in hydrology.In this study, the mash-up mode and the way of obtaining data required in mash-up application were analyzed. To obtain hydrological information from Internet,web crawler was designed to obtain news information about hydrology. The Open API,web service and screen-scraping technique were used to get data related to hydrology,including the water level in river and reservoir, river discharge, precipitation and water quality data. To use hydrological news in map mash-up, this paper focused on methods of extracting news elements from news page, including news text, keyword, abstract,date and place. The methods included extraction of news text based on text density,extraction of keywords based on information entropy, and extraction of place by using the term frequency and distribution. Based on the Naive Bayesian classifier, an improved method of text classification with keywords was proposed. It was verified that the method can improve the speed and accuracy of text classification.User interest model and personalized recommendation algorithm are the cores of personalized information service. In this study, representation method, building process and update mechanism of user interest model were proposed. Vector space mode was used to represent user interests. The user’s interested words were extracted from pages by analyzing user’s browsing history, and weights of these words were calculated according to browsing behavior. The Ebbinghaus Forgetting Curve was used to simulate the attenuation process of interests. Three kinds of commendation algorithms were designed based on the above user interest model, including content-based recommendation, user-based collaborative filtering recommendation and associationrules recommendation, and using process and applicable situation of each recommendation algorithm were also described. Considering the influence of geographical location on choices of users, geographical location factor was integrated in personalized algorithm to optimize the service result.On the basis of the research above, a B/S architecture service platform of personalized hydrological information aggregation was finally designed and realized to meet users’ demands to multi-source hydrological information and personalized information service.