Promises made by politicians, corporate leaders, and public figures have a significant impact on public perception, trust, and institutional reputation.However, the complexity and volume of such commitments, coupled with difficulties in verifying their fulfillment, necessitate innovative methods for assessing their credibility.This paper introduces the concept of Promise Verification, a systematic approach involving steps such as promise identification, evidence assessment, and the evaluation of timing for verification.We propose the first multilingual dataset, ML-Promise, which includes English, French, Chinese, Japanese, and Korean, aimed at facilitating in-depth verification of promises, particularly in the context of Environmental, Social, and Governance (ESG) reports.Given the growing emphasis on corporate environmental contributions, this dataset addresses the challenge of evaluating corporate promises, especially in light of practices like greenwashing.Our findings also explore textual and image-based baselines, with promising results from retrieval-augmented generation (RAG) approaches. 1

Text feedback from urban scenes is a crucial tool for pedestrians to understand surroundings, obstacles, and safe pathways. However, existing image captioning datasets often concentrate on the overall image description and lack detailed scene descriptions, overlooking features for pedestrians walking on urban streets. We developed a new dataset to assist pedestrians in urban scenes using 360-degree camera images. Through our dataset of Text360Nav, we aim to provide textual feedback from machinery visual perception such as 360-degree cameras to visually impaired individuals and distracted pedestrians navigating urban streets, including those engrossed in their smartphones while walking. In experiments, we combined our dataset with multimodal generative models and observed that models trained with our dataset can generate textual descriptions focusing on street objects and obstacles that are meaningful in urban scenes in both quantitative and qualitative analyses, thus supporting the effectiveness of our dataset for urban pedestrian navigation.

Aspect category sentiment analysis (ACSA) has attracted considerable research interest for its advanced analysis capabilities; however, it confronts challenges such as limited datasets and data imbalances. In this work, we introduce a novel model that addresses data imbalance by leveraging semantic knowledge from aspect category labels, thus enabling more nuanced sentiment analysis. Our analysis demonstrated superior performance compared to benchmark models in comparative experiments conducted on the SemEval 2015 and 2016 restaurant datasets. More specifically, a 3.97% increase in accuracy and a 7.25% rise in the F-1 score for the 2015 dataset were achieved, alongside a 0.94% improvement in accuracy and a 2.41% increase in the F-1 score for the 2016 dataset. Additionally, we utilized a novel data augmentation (DA) approach, combining back-translation and GPT-3 augmentation, to enhance the dataset for skiers’ ACSA. This enhancement aims to improve our model’s predictive accuracy on skiers’ reviews. The experimental results showed that the proposed model, trained on the augmented training set, exhibited an average increase of 0.41% in accuracy and 1.78% in the F-1 score. Lastly, a case study at Nozawa Onsen ski resort using a novel approach combining ACSA and IPA demonstrated the approach’s effectiveness in identifying important aspects and gauging skier satisfaction. This study highlighted that significant attention should be paid to skiers’ dissatisfaction with aspects such as the crowdedness of the slopes at the resort.

In natural language processing tasks, dictionaries must be built and words must be converted into their corresponding embeddings, creating problems such as out of vocabulary (OOV). To solve this problem, algorithms such as byte pair encoding (BPE) have been widely used to generate sub-words from known words and decompose OOV words into existing sub-words. However, Chinese has many types of characters called Kanji, and their distribution in the text is sparse, reducing the performance of BPE algorithms that rely on the frequency of occurrence. Compared with Kanji characters, radicals are less varied and more densely distributed. Moreover, because Kanji characters are hieroglyphs, their glyphs contain a rich amount of information and are unique. In this paper, we take advantage of these characteristics of Chinese and apply the BPE algorithm after converting the characters to the corresponding radicals, and combining sub-word embedding with the glyph feature vector. Experiments show that converting more Kanji characters to radicals results in a more even distribution of characters in the converted Chinese text, and the number of OOV words continues to decrease. Furthermore, at a 30% conversion ratio, the effect of radical-based embedding can outperform the effect of character-level embedding that keeps all Kanji characters unchanged. The results also indicate that combining with embedding using glyph features can improve performance despite the different conversion ratios.

To reduce the cost of administrative services, many local governments provide a frequently asked questions (FAQ) page on their websites that lists the questions received from local inhabitants with their official responses. The number of Q&A items posted on the FAQ page, however, will vary depending on the local government. To address this issue, we propose a method for augmenting local government FAQs by using a community-based Q&A (cQA) service. We also propose a new FAQ augmentation task to identify the regional dependence of Q&A to achieve the goal mentioned above. In our experiments, we fine-tuned the bidirectional encoder representations from transformers (BERT) model for this task, using a labeled local-government FAQ dataset. We found that the regional dependence of Q&As can be identified with high accuracy by using both the question and the answer as clues and with fine tuning for the deeper layers in BERT.

The growing ecosystem of Web APIs and IoT devices enables us to create a mashup within a short period of time. However, it is difficult to find appropriate Web APIs and IoT devices. Moreover, investigating a combination of Web APIs and IoT devices is much more difficult. The growing ecosystem enables us to achieve ideas in alternative ways, although it is difficult to find such alternative ways. We propose a recommendation system that takes the quality characteristics and such alternative ways into account. The system is implemented using a Steiner tree approximation algorithm, which can recommend a combination of Web APIs and IoT devices suitable for mashups. We tested the system in two aspects, i.e., the validity and performance of the algorithm. The validity was proved by running the system against a graph of actual Web APIs and IoT devices. The performance was proved by running the system against large graphs generated using a pseudorandom number generator. The system can output a result within 8 s against 28,200 nodes and five terminals, which is a considerably shorter time than that required by skilled developers to come up with a combination.

We propose a method to clarify the evolution of users' information needs related to a user's interests and actions based upon life events such as "childbirth." First, we extract topic transitions using dynamic topic models from blogs posted by users who have experienced life events. Next, we select the topics by computing the differences in topic probabilities before and after the life event. We evaluated our method based on three life events: "childbirth," "finding employment," and "marriage." Our method selected life event-relevant topics such as "child development," "working life," and "wedding ceremony." We found mothers' information needs such as "how to introduce baby food," employees' information needs such as "preparing an induction programme," and couples' information needs such as "wedding reception planning" in each topic.

Active collaboration with the public is the key to improving the administrative services of local government, and social media is an essential tool for understanding public sentiment. In this paper, we propose a method to gauge public sentiment for local government by using Twitter. We conducted an operational test with local government officers and found that our proposal was effective in revealing sentiments that were meaningful for the public, such as avoiding full parking lots, finding potholes in the road, or attracting favorable comments or suggestions to improve festival events.

We propose a method for finding impressive creators in online social network sites (SNSs). Many users are actively engaged in publishing their own works, sharing visual content on sites such as YouTube or Flickr. In this paper, we focus on the Japanese illustration-sharing SNS, Pixiv. We implement an illustrator search system based on user impression categories. The impressions of illustrators are estimated from clues in the crowdsourced social-tag annotations on their illustrations. We evaluated our system in terms of normalized discounted cumulative gain and found that using feedback on motifs and impressions for illustrations of relevant illustrators improved illustrator search by 11%.

Community Question Answering (CQA) sites such as Yahoo! Answers have emerged as rich knowledge resources for information seekers. However, answers posted to CQA sites can be irrelevant, incomplete, redundant, incorrect, biased, ill-formed or even abusive. Hence, automatic selection of "good" answers for a given posted question is a practical research problem that will help us manage the quality of accumulated knowledge. One way to evaluate answer selection systems for CQA would be to use the Best Answers (BAs) that are readily available from the CQA sites. However, BAs may be biased, and even if they are not, there may be other good answers besides BAs. To remedy these two problems, we propose system evaluation methods that involve multiple answer assessors and graded-relevance information retrieval metrics. Our main findings from experiments using the NTCIR-8 CQA task data are that, using our evaluation methods, (a) we can detect many substantial differences between systems that would have been overlooked by BA-based evaluation; and (b) we can better identify hard questions (i.e. those that are handled poorly by many systems and therefore require focussed investigation) compared to BAbased evaluation. We therefore argue that our approach is useful for building effective CQA answer selection systems despite the cost of manual answer assessments.

Opinion holder extraction research is important for discriminating between opinions that are viewed from different perspectives. In this paper, we describe our experience of participation in the NTCIR-6 Opinion Analysis Pilot Task by focusing on opinion holder extraction results in Japanese and English. Our approach to opinion holder extraction was based on the discrimination between author and authority viewpoints in opinionated sentences, and the evaluation results were fair with respect to the Japanese documents.

We propose a method to rate the of news articles using three clues: (1) commonality of the contents of articles among different news publishers; (2) numerical agreement versus contradiction of numerical values reported in the articles; and (3) objectivity based on subjective speculative phrases and news sources. We tested this method on news stories taken from seven different news sites on the Web. The average agreement between the system-produced credibility and the manual judgments of three human assessors on the 52 sample articles was 69.1%. The limitations of the current approach and future directions are discussed.