Extracting Fuzzy Linguistic Summaries Based On Including Degree Theory And FCA

With the popularity of computers and developing of internet, people are in face of huge number of information. The complexity of structure and relation is beyond people's understanding; In general information system or database, attributes domain are continuous numerical value or character, and it can not supply knowledge in these crude data; In addition, people often communicate with others using natural languages. Language knowledge can not found in data. So an artificial intelligence method is necessary to obtain the language knowledge.In this paper, extracting fuzzy linguistic summary is used to obtain useful knowledge in large number of data. It is very useful in many aspects. For example in market forecasting, decision making, art estimating, setting investing and so on.Discretization can accelerate learning and simplify concluding knowledge, even improving sort precision. Therefore information system with attributes of continuous domain must be discretized. Fuzzy discretization method is used in this paper. Each attribute (or named linguistic variable quantity) is divided into several linguistic summarizers. Triangular membership functions are used to define every linguistic summarizer in our work. The value of one object on a linguistic summarizer is its membership degree. Then original information system will be transformed into a fuzzy information system. Given a level value of cut set, the fuzzy information system will be transformed into a crisp information system.In obtaining crisp information, formal concept analysis is used to describe the relation between objects and linguistic summarizers. And including degree is used to show soft including degree between sets. For the general form of fuzzy linguistic summary, computing with words and fuzzy sets theory are used to decide linguistic summarizer, linguistic quantifier and truth degree. Then simple fuzzy linguistic summaries are extracted. Complex fuzzy linguistic summary are not combined with simple fuzzy linguistic summaries. Based on logical "and", logical "or" and logical implication, how to extract complex fuzzy linguistic summaries is discussed.In order to evaluate the extracting fuzzy linguistic summaries, truth degree, covering degree and so on are employed. The relation between simple with complex fuzzy linguistic summaries are also compared. Finally, we use an example to explain the following theory. And experiment shows that our method is effectively and significatively.