Research On Automatic Summarization System Oriented To Cigarette Quality Evaluation

With the increasing competition in the market, that companies want to be tenablein fierce competition product market is no easy task, but not impossible. Whatcompanies need to do is to grasp the quality firmly, and meet consumer expectationsby constantly improving the quality of products, so that the consumers can trust it.After joining the WTO momentum of modern tobacco companies is increasinglyfierce, however, China’s tobacco industry is also facing fierce competition in themarket and more severe challenges; the quality of cigarette products has become animportant factor in measuring competitiveness of enterprises in the tobacco. Thequality of cigarette product directly affects the economic benefits of tobaccocompanies， it also determines the height of the company’s future development in thelong run. Because of this，each tobacco enterprise puts quality attaches to the greatimportance position. Up to now, the merits and demerits of the quality of cigaretteshas been evaluated by the sense of evaluating experts. So the tobacco companies willorganize evaluating experts smoke panel test in the development of a new product orproduct improvements, and cigarette evaluating experts will evaluate the quality ofthe product and feedback. And a large number of the texts of quality evaluation needhuman analysis and sorting, it consumes relatively long time, and prone to error. Inthis paper, cigarette-oriented quality evaluation system of automatic summarization isa very effective tool to solve this problem.Modern society is an era of information explosion; information overload hasbecome a significant problem. Traditional information retrieval methods can not meetpeople’s demands to get great information, the summary as compressed information ofthe original text can reduce the amount of information. People can use computerpreprocessing a lot of text messages, and generate summary information that canbasically reflect the main content of the article, and then can make roughly the judgment through reading only a small amount of information. If we are interested,we can read the full text further, and if we want to summary basic information about adocument, we can get a rough summary of the content, which will greatly improveefficiency that people get electronic text message. By reading more accurate summaryinformation, we can quickly and easily understand the text without having to readthrough the entire document, thus can save valuable time and effort. Automaticsummarization is the topic that linguistic intelligence and computer science areconcerning together, its essence is information concentrating and Information mining.In theory, the study of automatic summarization will help people derive knowledgemodel to explore knowledge, and generalize, understand natural language text,automatic summarization is also considered an important symbol that computerrealize natural language understanding. From the application point of view, on theInternet and the rapid development of electronic literature today, the use of automaticsummarization system will significantly reduce the cost of artificially preparedsummaries, shorten the publication period of abstracts, and provide people with therequired information that can obtain quickly, accurately and inexpensively.In this paper, the automatic summarization system oriented to cigarette qualityevaluation takes the method that it automatically extracts the sentences from theoriginal text to generate summarization. The system is developed with.NTEFramework and SQL Server2005.The system directly makes use of computerautomatically generate text summarization in tobacco fields, its function is that thedocumentation about quality evaluation of cigarette products generate summary oftext message by statistical analysis and sentiment analysis and output according tocertain standard. In this paper, the system introduces the specific implementation ofeach functional module in detail, it involves five modules. Text preprocessing modulepreprocess text information, the text can standardize according to certain rules;segmentation module, it completes Chinese word segmentation in text message, thetechnology of Chinese automatic segmentation is a very important basis work in thefield of natural language processing; loading the thesaurus module makes Word thesaurus more and more rich, increasing the accuracy of segmentation. Wordfrequency statistics and analysis module, finds the keyword by statisticsing nounsindicators and adjective sentiment words, and prepares a summary for the extraction.Emotional polarity judgment module, is calculated by a weighted sum of keywords inthe text message collection belongs sentence adjective polarity, and obtain emotionalpolarity that the key words possesses. Summary output module, according to theabove analysis module, gets summary information in accordance with certain rules,thus achieve the function of user’s needs.In this paper, the system solves the needs of users, can greatly improve thequality and efficiency of test and analysis personnel, reduces errors and labor intensity,and thereby increases rapidly product quality and overall management of tobaccocigarette enterprises. The system has a certain practicality and promotional value.