Information Engineering Express

Flaming Participants Detection Using Account and Stylistic Characteristics from SNS

Linshuo Yang — 2024-08-03

Recent years have seen a rapid increase in social media network (SNS) users due to their swift growth. Consequently, people can easily engage in interactive communication with a vast and undefined audience. This has given rise to a recurring phenomenon called “flaming”, in which critical comments flood SNS. While various studies on flaming have been conducted, most of them have primarily focused on individuals receiving significant volumes of critical comments, rather than those who compose them, referred to as “flaming participants”. In this study, we examine the characteristics of flaming participants on Twitter (Although the name has now been changed to “X”, this paper still uses “Twitter” as its name) by using machine learn-ing to classify them into two groups: flaming participants and normal users. For the classification features, we utilize account information, i.e., statistical data for each account, and stylistic features of the postings, i.e., (1, n)-grams of the part-of-speech tags of the postings. Our experimental findings underscore the effectiveness of these features in identifying Twitter’s flaming participants. Additionally, our research reveals that flaming participants tend to employ quote tweets more frequently than typical users, and there are distinctive word patterns observable among flaming participants.

Generating a Technical Trend Map by Analyzing the Structure of U.S. Patents Using Patent Families

Jun Nakamitsu — 2024-08-12

Researchers and developers search for patents in fields related to their own research to obtain information on issues and effective technologies in those fields for use in their research. However, it is impossible to read through the full text of many patents, so a method that enables patent information to be grasped briefly is needed. In this study, we analyze the structure of U.S. patents with the aim of extracting important information. Using Japanese patents with structural tags such as “field”, “problem”, “solution”, and “effect”, and corresponding U.S. patents (patent families), we automatically created a dataset of 81,405 U.S. patents with structural tags. Furthermore, using this dataset, we conduct an experiment to assign structural tags to each sentence in the U.S. patents automatically. For the embedding layer, we use a language representation model BERT pretrained on patent documents and construct a multi-label classifier that classifies a given sentence into one of four categories: “field”, “problem”, “solution”, or “effect”. We are able to classify sentences with precision of 0.6994, recall of 0.8291, and F-measure of 0.7426. We have analyzed the structure of U.S. patents using our method and generated a technological trend map, which confirms the effectiveness of the proposed method.