The Research And Application Of Multi-Dimensional Sequential Pattern In The Analysis Of Broadcast Listen Rate

With the broad applications of the database and the Internet, we have the more and more data. In the field of the survey of broadcast listen data, Digitizing the broadcast media became current. Submerged by data, we are hungry for knowledge. Data Mining can solve this problem. Now in the field of survey of broadcasting from 1930s, the Nielsen company come into existence and BBC have begun to do this survey everyday. The analysis and description of broadcasting's effect is substituted by digital. Because of lacking of competence of re-analyses and no good at anatomizing the data of tuning rate, these companies only passively took the superficial data presented by the surveying organization. For the problem of survey of broadcasting in this field, the dissertation expatiates that the algorithm of multi-dimensional sequential pattern is applied in analyzing of data of survey of broadcasting. This method can improve the information's management for broadcasting station by based on index data and information of data mining. Also it can be more scientific used in program estimating, personnel assessing and analyzing of audience etc. At the same time, it provides the valuable data for the professional studies of broadcasting media'domain.Based on some classical algorithms of multi-dimensional sequential pattern, the dissertation introduces a new algorithm that based, that also takes the existing problem of broadcasting tuning and the professional knowledge into consideration. The algorithm firstly takes multi-dimensional'information as a element and inserts it into sequence, forming a new sequence. Then based-PrefixSpan algorithm, and applying the technologies of timearea's merger and spreading, the new algorithm mines the new multi-dimensional sequential pattern with timer.Taking Zhongshi GuangLian's listen rate analyzing system as background, after achieving computing the elementary index of broadcasting listen rate, based-tense contained multi-dimensional sequential pattern algorithm has been used in it. The method is able to complish the process of data mining, from the pretreatment of rude data to the estimating and optimizing of mining outcome in finally.After summarizing the dissertation, it gives the further direction and ideas.