نوع مقاله : مقاله پژوهشی
نویسنده
استادیار، گروه مهندسی کامپیوتر، واحد تهران شمال، دانشگاه آزاد اسلامی، تهران، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسنده [English]
Purpose: The question of what methods have been used by researchers to solve a given scientific problem is an important question for a researcher who intends to address that scientific problem. Important concepts and methods in any field of science are mentioned in the technical terms of that field. Therefore, the "usage" relationship between technical terms is of considerable importance. The purpose of this research is to establish a method that recognizes the usage relationship between terms present in the titles of scientific articles and to include those terms and the usage relationship between them in an ontology. Identifying and recording the usage relationship between terms, when implemented on a large scale, reveals what approaches are examined by researchers to address a given task.
Method: This research is applied in purpose and qualitative in nature. The primary data are the titles of articles in a selected scientific journal, a specified time period. We propose a method which detects technical terms present in the title of scientific articles, finds two relations between those terms, namely the hyponym-hypernym and usage relations, and inserts the result into an ontology. First, the noun phrases present in the title of the article are detected. These noun phrases are the technical terms, with some reservations. Our method focuses on titles in which two technical terms are related through a connector, which semantically indicates usage. Such cases are inserted into the ontology after detection. In addition to the usage relation, a hyponymy relation is also proposed based solely on the syntactic structure of each found noun phrase. Appropriate inference rules are designed in the ontology, based on which the usage relation between terms is inherited from the hyponym to the hypernym. Although the usage relationship between the article title components can also be extracted using large language models, performing this for a large number of article titles is costly. In addition, it will be necessary to analyze the response of the language models. Also, the need for inclusion in the ontology will remain.
Findings: To evaluate the method, the titles of one year of articles from a scientific journal were used. After implementing the method, the ontology was examined. In 69% of the titles which contained the pattern of this study, both noun phrase, i.e. the used term and the using term, were completely extracted. In 17% of the titles with the pattern, one or both noun phrase, i.e. at least one of the two terms, were incompletely extracted, so that the extracted part is semantically correct but does not contain the entire meaning of the title. In 14% of the titles with the pattern, at least one of the two noun phrase, i.e., the technical used term or the using term, was extracted incorrectly.
Conclusion: Since the proposed method is automatic, it can be applied to a large number of scientific articles. To increase accuracy, it is suggested that other connectives be considered in the analysis of the article title, in addition to the connectives that reflect the usage relationship.
کلیدواژهها [English]
)
فرستادن نظر درباره این مقاله