International Journal of Service and Knowledge Management

Interpretable Document-Level Polarity Classification with Inter-sentence Attention

Daisuke Ikeda — 2026-02-25

Large language models (LLMs), such as BERT and GPT-3, significantly impact our daily lives. They can derive accurate outputs for many tasks about natural languages, but they do not explain the reason for the outputs. In this sense, LLMs work as a black box and have the problem of interpretability. This paper is devoted to considering polarity classification of documents. Compared to simple sentence-level polarity classification, document-level one makes it more difficult to create a high-performance model because we have to consider relationships between sentences. To tackle this change, we use inter-sentence attention, which can capture the relationship between sentences: the higher an inter-sentence attention score is, the more relevant the corresponding sentences are to each other. We use intersentence attention scores to capture the context of sentences and propose a model whose classification is more similar to human judgment. To validate the proposed model, we conduct three types of experiments: one is to compare classification performance with prior models; the second one is to compare interpretability with prior models; and the last one is to show the ability of inter-sentence attention whether it could capture the relationship between sentences. From the first two types, we found that our model is more accurate on two real datasets. In the second type of experiment to assess interpretability, we examined the overlap between sentences that contribute to the model’s predictions and those annotated by humans for the same document and found that our model has a larger overlap and is more likely to extract interpretive sentences that humans intuitively consider important. In addition, our result partially captures the polarity of “implicit” sentences that do not contain direct expressions, which could not be captured by prior models, suggesting that our model may lead to a more natural interpretation. From the third type of experiment, we show that our model can capture the contexts of sentences.

Motivations for Migration: Why Migrant Entrepreneurs Choose to Settle in Depopulated Communities?

Osamu Nakajima — 2026-02-25

This study uses Higashikawa Town, Hokkaido, as a case study, to investigate the motivations of migrant entrepreneurs to migrate to depopulated rural areas and settle there for the long term. As the population flows out to urban areas and the population ages, depopulation poses serious challenges to the maintenance of local industries and communities. In this study, the authors conducted semi-structured interviews with six migrant entrepreneurs and clarified the motivations for migration and the factors behind them through co-occurrence network analysis using KH Coder. As a result, the authors found that migrant entrepreneurs often migrate in search of self-realization and professional independence, which are largely contributed to by local networks, government support, and cultural and economic assets such as the ''Craft Road''. These supports and resources not only increase the feasibility of achieving individual goals, but also promote cultural exchange and regional revitalization. This study also showed that meeting the needs of migrant entrepreneurs and supporting them in achieving their individual goals is an important factor in realizing long-term settlement. It is believed that these supports and resources play a role in promoting cultural exchange and regional revitalization as well as enhancing regional sustainability.

Strategies for Conventions Bidding in a Competitive Environment

Hidekazu Iwamoto — 2026-02-25

This study identifies key factors that Japanese MICE stakeholders believe to contribute to successful bids for conventions in competitive environments. MICE has a large economic ripple effect on the local economy. However, there are few in-depth case studies of successful convention bids in Japan. An analysis of a questionnaire survey of 220 stakeholders from two IME events identified five factors: (1) the amount of information available in English, (2) cooperative stakeholder structures, (3) hotel brand recognition, (4) use of convention facilities, and (5) cultural and tourism resources. These findings emphasize the importance of infrastructure, stakeholders’ collaboration, and cultural appeal. The study is limited by its small sample size and the need for more research.