Seiji Tsuchiya

Single Document Summarization Based on Local Topic Identification and Word Frequency

In this task, an approach for single document summaries based on local topic identification and word frequency is proposed. In recent years, there has been increased interest in automatic summarization. The physical features are often used and have been successfully applied to this field; it also has some disadvantages of non-redundancy, structure and coherence. Therefore, we introduced logical structure feature which has been successfully applied in multi-document summarization (MDS), and we designed a system to accomplish this task. Documents can be clustered into local topic after sentences similarity is calculated, which can be sorted by the scoring. Then sentences from all local topics are selected by computing the word frequency. Using this proposed method, the information redundancy of each local topic and among local topic is reduced. The information coverage ratio and structure of the summarization is improved.

Opinion mining: A study on semantic orientation analysis for online document

As a result of advance in technology, there now exist a large amount of online documents in the form of surveys or called reviews. Most of the previous work on text classification is focusing on sentiment text classification. Sentiment classification requires the knowledge data of vocabulariespsila semantic meaning and the relationships between the vocabularies. In this paper, sentiment features of text were divided into characteristic words and phrases, which were extracted from the training data. The method combining HowNet with sentiment classifier was proposed. It computed semantic similarity of characteristic words, phrases with tagged words in HowNet, and it adopted the positive and negative terms as features of sentiment classifier. In the test, a sentiment classifier was designed to compare with the other methods. Evaluation results show the effectiveness of our method.

Emotion estimation algorithm based on interpersonal emotion included in emotional dialogue sentences

Emotion recognition aims to make computer understand ambiguous information of human emotion. Recently, research of emotion recognition is actively progressing in various fields such as natural language processing, speech signal processing, image data processing or brain wave analysis. We propose a method to recognize emotion in dialogue text by using originally created Emotion Word Dictionary. The words in the dictionary are weighted according to the occurrence rates in the existing emotion expression dictionary. We also propose a method to judge the object of emotion and emotion expressivity in dialogue sentences. The experiment using 1,190 sentences proved about 80% accuracy.

Classification of Facemarks Using N-gram

In this paper, we present an approach for the classification of facemarks into some facial expression categories. Facemarks are some of the expressions that are often used in text-based communication. Facemarks express human facial expression or action and they help us to understand what writers imply. However, there are some problems in the proccessing of the facemarks with computer; facemarks are numerous and users make new facemarks. Therefore, we propose to use the characters in facemarks to classify them. Though the facemarks are various, only some of signs and letters are used for the characters that compose the them. Moreover, there is a feature of the expressions of facemarks in the characters. We present an approach for facemarks classification using N-gram and evaluate this method.

A sensuous association method using an association mechanism for natural machine conversation

Humans manipulate smooth communications by retrieving, understanding and judging the sensuous characteristics from conversations consciously or unconsciously. This paper presents sensuous association methods to judge noun association from combinations of nouns (or adjectives) and adjectives/adjectival equivalents expressing senses as a means of achieving sense judgment similar to the common-sense judgment made by humans. The propose method is based on a mechanism that makes it possible to associate various concepts from a given concept. The precision rate and recall rate for nouns associated from combinations of nouns (or adjectives) and adjectives/adjectival equivalents were approximately 96.3% and 63.6% in reference to human judgment.

Human emotion model based on discourse sentence for expression generation of conversation agent

There was a conversation agent on the generation method of facial expression. It is necessary for the conversation system like the human for communication. In the previous method, at first a word which could influence the feeling was defined. Facial expression was changed according to the word which influences the feeling in discourse. Whereas, facial expression could not be changed if there was not a word that was defined in the discourse. Hence, we proposed a human emotion model for the expression generation of the conversation agent. The method based on the human emotion model can solve problem of the previous method and may make a more humanity conversation agent. In this study, we put a human emotion tag to discourse of talks scenarios and model to conversation agent of human emotion. There were two kinds of methods that put the human emotion tag to discourse. We make the human emotion model by scenarios that adopts the human emotion tag. Used the human emotion model to create facial expression of conversation agent. The assessment experiment was performed by using the systems of previous method and two human emotion models, and compared the results between the three methods.

A time judgement system based on an association mechanism

Common sense and judgement ability, in the same way as humans, are necessary to realize a computer that communicates directly with humans. Such a thing needs the ability to recall a concept from a particular word and to associate the concept with others. A Time Judgement System that can understand everyday time expressions is important for natural communication, and the system was based on the aforesaid Association Mechanism. The purpose of this research is to construct the system that can treat regular time expressions with adaptability to unknown expressions. The resultant Time Judgement System achieves a correct response rate of 75.9% with accuracy of 85.8%, comparable to that of human subjects.

Emotion judgment method from an utterance sentence

Authors focus on the emotion of such common sense and attempt to establish a method to judge the users emotions based on utterances. A speakers utterance sentence includes a linguistic proposition and a linguistic modality. The linguistic modality is an important factor to represent emotions for an utterance sentence. Therefore, in this paper, a method is proposed which judges speakers emotions from the linguistic proposition and modality included in the utterance sentence. The proposed method uses knowledge base and an Association Mechanism. As a result, the accuracy of the proposed emotion judgment method processing the linguistic modality is improved approximately 30% compared with an existing method.

Exploring Words with Semantic Correlations from Chinese Wikipedia

In this paper, we work on semantic correlation between Chinese words based on Wikipedia documents. A corpus with about 50,000 structured documents is generated from Wikipedia pages. Then considering of hyper-links, text overlaps and word frequency, about 300,000 word pairs with semantic correlations are explored from these documents. We roughly measure the degree of semantic correlations and find groups with tight semantic correlations by self clustering.

A Semantic Information Retrieval Technique and an Evaluation for a Narrow Display Based on an Association Mechanism

The purpose of this study is to output information which users hope as high ranking results on narrow display of mobile information terminals. The concept of the proposed algorithm is to understand the meanings and the contents expressed by keywords and to retrieve appropriately related information. Concretely, we propose an information retrieval technique to evaluate the relationships between words using an association mechanism. A new test collection was made by 100 examinees who judged the retrieval results according to text meanings. Retrieval performance was objectively confirmed with the experiment based on the test collection.