Yoshiaki Kurosawa

A classification method for spam e-mail by Self-Organizing Map and automatically defined groups

We have some difficulties in E-mails as a communication tool, because the number of E-mails infected with virus and/or recognized as Spam increases. Some E-mail filter softwares removes such problematic ones. However, we may mett the misjudgements for the filtering the Spam E-mail, even if the E-mail is important and then we cannot receive it. In this paper, we propose a classification method for Spam E-mail based on the results of SpamAssassin, which is the open source software to identify spam signatures. This method can learn patterns of Spam E-mails and Ham ones and correctly recognize them. First, the method divides E- mails into some categories by Self-Organizing Map(SOM) and extracts the correct judgement rules by Automatically Defined Groups(ADGs), even if the results by SpamAssassin are wrong. In order to verify the effectiveness of our proposed method, we examined approximately 3,000 E-mails.

A Description Method of Syntactic Rules on Japanese Filmscript

The purpose of this paper is to propose a description method of syntactic rules, particularly, to manage manually. Our aims are the decrease of the number of rules and the structuring of them. The former is achieved by describing the rules with regular expression, and the latter is constructed by step-by-step analysis. These descriptions make the management of the rules simple and easy. Considering such concerns, we report the effectiveness of our method with some experimental results.

A method using acoustic features to detect inadequate utterances in medical communication

We previously proposed a method that uses grammatical features to detect inadequate utterances of doctors. However, nonverbal information such as that conveyed by gestures, facial expression, and tone of voice are also important. In this paper, we propose a method that uses eight acoustic features to detect three types of mental states (sincerity, confidence, and doubtfulness/acceptance). A Support Vector Machine (SVM) is used to learn these features. Experiments showed that the systems accuracy and recall rates respectively ranged from 0.79-0.91 and 0.80-0.94.

A question-and-answer classification technique for constructing and managing spoken dialog system

To recognize user speech accurately and respond to it appropriately, a spoken dialog system usually uses a question-and-answer database (QADB) which contains many question-and-answer pairs. The systems first select a question example which is the most similar to the recognition result for the input voice from the database. An answer sentence which is then paired with the selected question example is output to the user. Many systems have a large database to enable a more appropriate answer to be output. However, when such a database is used, the waiting time increases because the system needs to find the most appropriate question example from a vast number of question examples. We propose a method of classifying the queries in the QADB. By classifying question examples into some clusters using pLSA, an appropriate question example can be found more quickly than when using the conventional method. We evaluated the validity of our proposed method by changing various parameters.

Automatically Defined Groups for Knowledge Acquisition from Computer Logs and Its Extension for Adaptive Agent Size

Recently, a large amount of data is stored in databases through the advance of computer and network environments. To acquire knowledge from the databases is important for analyses of the present condition of the systems and for predictions of coming incidents. The log file is one of the databases stored automatically in computer systems. Unexpected incidents such as system troubles as well as the histories of daily service programs’ actions are recorded in the log files. System administrators have to check the messages in the log files in order to analyze the present condition of the systems. However, the descriptions of the messages are written in various formats according to the kinds of service programs and application software. It may be difficult to understand the meaning of the messages without the manuals or specifications. Moreover, the log files become enormous, and important messages are liable to mingle with a lot of insignificant messages. Therefore, checking the log files is a troublesome task for administrators. Log monitoring tools such as SWATCH [1], in which regular expressions for representing problematic phrases are used for pattern matching, are effective for detecting well-known typical error messages. However, various programs running in the systems may be open source software or software companies’ products, and they may have been newly developed or upgraded recently. Therefore, it is impossible to detect all the problematic messages by the predefined rules. In addition, in order to cope with illegal use by hackers, it is important to detect unusual behavior such as the start of the unsupposed service program, even if the message does not correspond to the error message. To realize this system, the error-detection rules depending on the environment of the systems should be acquired adaptively by means of evolution or learning. Genetic programming (GP) [2] is one of the evolutionary computation methods, and it can optimize the tree structural programs. Much research on extracting rules from databases by GP has been done in recent years. In the research [3–5],

Emotion Oriented Intelligent System for Elderly People

We propose the “emotion oriented intelligent interface for elderly people” to be able to access computers easily. We applied three methods about natural language dialogue and emotion, analyzing the user’s utterances, estimating and expressing the user’s emotions, and analyzing the user’s intention from his/her utterances. By using these three methods, the user would be able to communicate with the system naturally. We constructed an interface system based on the methods, and the interface system has been applied into the “web-based health service system for elderly people.”

Natural language dialog system considering speaker’s emotion for open-ended conversation

To respond appropriately to an utterance, human-like communication system, should consider not only words in the utterance but also the speaker’s emotion. We thus proposed a natural language dialog system that can estimate the user’s emotion from utterances and respond on the basis of the estimated emotion. To estimate a speaker’s emotion (positive, negative, or neutral), 384 acoustic features extracted from an utterance are utilized by a Support Vector Machine (SVM). Artificial Intelligence Markup Language (AIML)-based response generating rules are expanded so that the speaker’s emotion can be considered as a condition of these rules. Two experiments were carried out to compare impressions of a dialog agent that considered emotion (proposed system) with those of an agent that did not (previous system). In the first experiment, 10 subjects evaluated the impressions after watch four conversation videos (no emotion estimation, correct emotion estimation, inadequate emotion estimation, and imperfect emotion ...

An Intelligent Maintenance System with Open Source Software

Recent computer systems are popularly composed of two or more computer servers connected via LAN (Local Area Network), where each computer server individually serves a specific function. The computer system provides the required service via the combination of individual server operations. When running such networked systems, a system administrator must monitor the occurrences of errors. However, there are often a lack of engineers who have expert knowledge in all fields related to the system, including hardware and software knowledge. Each computer is equipped with various hardware components, such as a CPU, memory, and hard disk drives. If any one of such parts breaks down, the computer will not operate. When the faulty component carries out various functions within the computer, it is difficult to specify which part is broken. Moreover, it is difficult to detect whether the state of the total system is stable or not, because various software applications work in each computer and many computers connected via LAN cooperate to realize the total service. In this chapter, we propose a web-based management tool for maintaining computers in a LAN environment. The proposed tool has two main features. One function is to detect hardware faults by utilizing Cacti [1]. Cacti is a complete network graphing solution designed to harness the power of RRDTool (Round Robin Database Tool)’s data storage and graphing functionality. The other function is to detect software errors from abnormal state messages appearing in system LOG files not only in the operating system (OS) but also in the application software. The system

An error measure for japanese morphological analysis using similarity measures

The aim of this paper is to propose a Japanese morphological error measure in order to automatically detect morphological errors when analyzing. In particular, we focused on three similarities as the measure and experimented using them. From our experimental results, it was found that the precision of one of measures was 74% and it functioned well.

Natural Language Dialog System Considering Speaker’s Emotion Calculated from Acoustic Features

With the development of Interactive Voice Response (IVR) systems , people can not only operate computer systems through task-oriented conversation but also enjoy non-task-oriented conversation with the computer. When an IVR system generates a response, it usually refers to just verbal information of the user’s utterance. However, when a person gloomily says “I’m fine,” people will respond not by saying “That’s wonderful” but “Really?” or “Are you OK?” because we can consider both verbal and non-verbal information such as tone of voice, facial expressions, gestures, and so on. In this article, we propose an intelligent IVR system that considers not only verbal but also non-verbal information. To estimate a speaker’s emotion (positive, negative, or neutral), 384 acoustic features extracted from the speaker’s utterance are utilized to machine learning (SVM). Artificial Intelligence Markup Language (AIML)-based response generating rules are expanded to be able to consider the speaker’s emotion. As a result of the experiment, subjects felt that the proposed dialog system was more likable, enjoyable, and did not give machine-like reactions.