AI\u306b\u6587\u5b66\u4f5c\u54c1\u306e\u611f\u60c5\u3092\u8a55\u4fa1\u3055\u305b\u308b\u3068\u304d\u3001\u70b9\u6570\u306e\u9055\u3044\u306f\u4f55\u304b\u3089\u751f\u307e\u308c\u308b\u306e\u3067\u3057\u3087\u3046\u304b\u3002\u672c\u7814\u7a76\u3067\u306f\u3001AI\u306b\u6307\u793a\u3059\u308b\u5f79\u5272\u3084\u8aad\u307f\u306e\u7acb\u5834\u3067\u3042\u308b\u300c\u30da\u30eb\u30bd\u30ca\u300d\u3068\u3001\u751f\u6210\u6642\u306e\u9078\u629e\u306e\u3070\u3089\u3064\u304d\u306b\u95a2\u308f\u308b\u8a2d\u5b9a\u300ctemperature\u300d\u306b\u6ce8\u76ee\u3057\u307e\u3057\u305f\u3002\u7279\u5b9a\u306e\u30da\u30eb\u30bd\u30ca\u3068temperature\u3092\u4e00\u5bfe\u4e00\u306b\u7d44\u307f\u5408\u308f\u305b\u308b\u3060\u3051\u3067\u306f\u3001\u3069\u3061\u3089\u306e\u8a2d\u5b9a\u306b\u5bfe\u5fdc\u3059\u308b\u9055\u3044\u306a\u306e\u304b\u3092\u533a\u5225\u3067\u304d\u307e\u305b\u3093\u3002 <\/p>\n\n\n\n
\u305d\u3053\u3067\u30016\u7a2e\u985e\u306eLLM\u306b\u65e5\u672c\u8a9e\u306e\u6587\u5b66\u4f5c\u54c13\u7de8\u3092\u63d0\u793a\u3057\u3001\u5927\u5b661\u5e74\u751f\u3001\u6587\u5b66\u7814\u7a76\u8005\u3001\u611f\u60c5\u8c4a\u304b\u306a\u8a69\u4eba\u3001\u611f\u60c5\u3092\u6301\u305f\u306a\u3044\u30ed\u30dc\u30c3\u30c8\u3068\u3044\u30464\u3064\u306e\u7acb\u5834\u3068\u30014\u6bb5\u968e\u306etemperature\u3092\u3059\u3079\u3066\u7d44\u307f\u5408\u308f\u305b\u307e\u3057\u305f\u3002\u5404\u6761\u4ef6\u306b\u3064\u304d5\u4ef6\u3001\u8a081,440\u4ef6\u306e\u5fdc\u7b54\u304b\u3089\u3001\u300c\u8208\u5473\u300d\u300c\u9a5a\u304d\u300d\u300c\u60b2\u3057\u307f\u300d\u300c\u6012\u308a\u300d\u30920\uff5e100\u3067\u8868\u3059\u8a085,760\u4ef6\u306e\u30b9\u30b3\u30a2\u3092\u5206\u6790\u3057\u307e\u3057\u305f\u3002\u00a0<\/p>\n\n\n\n
\u4eca\u56de\u306e\u6761\u4ef6\u3067\u306f\u3001\u8a2d\u5b9a\u306b\u3088\u308b\u5e73\u5747\u30b9\u30b3\u30a2\u306e\u9055\u3044\u306f\u4e3b\u306b\u30da\u30eb\u30bd\u30ca\u306b\u5bfe\u5fdc\u3057\u3001temperature\u5358\u72ec\u306b\u5bfe\u5fdc\u3059\u308b\u9055\u3044\u306f\u6bd4\u8f03\u7684\u5c0f\u3055\u3044\u7d50\u679c\u3068\u306a\u308a\u307e\u3057\u305f\u3002\u00a0 \u56f3: \u672c\u7814\u7a76\u3067\u306f\u3001AI\u306b\u4e0e\u3048\u308b\u5f79\u5272\uff08\u30da\u30eb\u30bd\u30ca\uff09\u3068\u751f\u6210\u8a2d\u5b9a\uff08temperature\uff09\u3092\u72ec\u7acb\u306b\u5909\u3048\u3001\u6587\u5b66\u4f5c\u54c1\u306b\u5bfe\u3059\u308b\u611f\u60c5\u8a55\u4fa1\u304c\u3069\u306e\u3088\u3046\u306b\u5909\u308f\u308b\u304b\u3092\u5206\u6790\u3057\u305f\u3002 <\/p>\n\n\n\n \u79d1\u7814\u8cbb\u7814\u7a76\u300cEQ-Fuzzy\u300d\u306e\u9032\u5c55\u3078 <\/strong><\/p>\n\n\n\n \u672c\u7814\u7a76\u306f\u3001\u79d1\u7814\u8cbb\u57fa\u76e4\u7814\u7a76\uff08C\uff09\u300c\u30d5\u30a1\u30b8\u30a3\u30a8\u30f3\u30c8\u30ed\u30d4\u30fc\u3068\u30da\u30eb\u30bd\u30ca\u591a\u69d8\u6027\u306b\u57fa\u3065\u304fLLM\u611f\u60c5\u77e5\u80fd\u306e\u4e0d\u78ba\u5b9f\u6027\u8a55\u4fa1\u300d\uff08\u8ab2\u984c\u756a\u53f726K14983\uff09\u3067\u9032\u3081\u308bEQ-Fuzzy\u7814\u7a76\u306e\u4e00\u74b0\u3067\u3059\u3002\u00a0 \u767a\u8868\u8ad6\u6587\u30fb\u7814\u7a76\u652f\u63f4 <\/strong><\/p>\n\n\n\n \u8ad6\u6587\u984c\u76ee<\/strong>: A Factorial Design for Separating Persona and Temperature Effects in LLM-Based Fuzzy Emotion Scoring <\/p>\n\n\n\n \u8457\u8005<\/strong>: Naruki Shirahama\uff08\u767d\u6ff5\u6210\u5e0c\uff09\u3001Yuma Yoshimoto\u3001Naofumi Nakaya\u3001Satoshi Watanabe <\/p>\n\n\n\n \u767a\u8868<\/strong>: SCIS&ISIS 2026\u30012026\u5e7411\u67083\u65e5\u3001\u6771\u4eac\u90fd\u516b\u738b\u5b50\u5e02\uff08Paper ID: 171\u3001Full paper\uff09 <\/p>\n\n\n\n \u7814\u7a76\u652f\u63f4<\/strong>: \u672c\u7814\u7a76\u306f\u3001JSPS\u79d1\u7814\u8cbbJP26K14983<\/a>\u304a\u3088\u3073JP24K13350<\/a>\u306e\u52a9\u6210\u3092\u4e00\u90e8\u53d7\u3051\u3066\u5b9f\u65bd\u3055\u308c\u307e\u3057\u305f\u3002 <\/p>\n\n\n\n <\/p>\n\n\n\n How Do AI\u2019s Emotion Ratings Change with Its \u201cRole\u201d and \u201cGeneration Settings\u201d?<\/strong> <\/p>\n\n\n\n Professor Shirahama presents at SCIS&ISIS 2026 \u2014 four faculty members from the Faculty of Data Science present their research<\/strong> <\/p>\n\n\n\n On 3 November 2026, Professor Naruki Shirahama of the Faculty of Data Science at Shimonoseki City University presented research on emotion ratings generated by large language models (LLMs) at SCIS&ISIS 2026, an international conference held in Hachioji, Tokyo. The study examined how these ratings changed when the reading perspective assigned to the AI and its response-generation settings were varied independently. <\/p>\n\n\n\n Separating the Effects of Roles and Generation Settings<\/strong> <\/p>\n\n\n\n When AI is asked to rate emotions in literary works, what accounts for differences in its scores? This study focused on two factors: the persona\u2014the role or reading perspective assigned to the AI\u2014and temperature, a setting that influences variability in the choices made during response generation. When personas and temperature settings are paired one-to-one, it is not possible to distinguish which setting is associated with the observed differences. <\/p>\n\n\n\n To address this issue, the researchers presented three Japanese literary works to six LLMs and tested every combination of four personas\u2014a first-year university student, a literary scholar, an emotionally expressive poet, and an emotionless robot\u2014and four temperature levels. With five responses per condition, they analyzed a total of 1,440 responses containing 5,760 scores for interest, surprise, sadness, and anger, each rated on a scale from 0 to 100. <\/p>\n\n\n\n Mean Scores and Fuzzy Entropy Show Different Patterns\u00a0<\/strong> Fuzzy entropy, meanwhile, quantifies the overlap in a score\u2019s membership of the categories \u201clow\u201d, \u201cmedium\u201d, and \u201chigh\u201d. In the analysis of this measure, the interaction between the two settings accounted for a larger share of variation than it did in the analysis of mean scores. This pattern was particularly evident for anger and surprise. <\/p>\n\n\n\n These findings highlight the importance of varying the two settings independently and interpreting mean scores separately from derived measures. However, the measure used here does not directly assess AI\u2019s subjective feelings or human psychological states, and the conclusions are limited to the models and literary works examined in this study. <\/p>\n\n\n\n Figure: The study independently varied the roles (personas) assigned to AI and the generation setting (temperature) to examine changes in emotion scores for literary texts. <\/p>\n\n\n\n Advancing the KAKENHI-Supported EQ-Fuzzy Project <\/strong><\/p>\n\n\n\n This study forms part of EQ-Fuzzy, a project supported by a JSPS KAKENHI Grant-in-Aid for Scientific Research (C) (Grant Number 26K14983). The project focuses on evaluating uncertainty in LLM emotional intelligence using fuzzy entropy and persona diversity. <\/p>\n\n\n\n The discussion following the presentation covered comparisons with human ratings, potential applications in dialogue with robots, and AI responses to particular emotions. How these ratings relate to human evaluations, and whether similar findings extend to other literary works and languages, remain questions for future research. <\/p>\n\n\n\n In addition to Professor Shirahama, three other faculty members from the Faculty of Data Science attended the conference, each presenting their own research. <\/p>\n\n\n\n Paper and Research Support <\/strong><\/p>\n\n\n\n Paper title<\/strong>: A Factorial Design for Separating Persona and Temperature Effects in LLM-Based Fuzzy Emotion Scoring <\/p>\n\n\n\n Authors<\/strong>: Naruki Shirahama, Yuma Yoshimoto, Naofumi Nakaya, and Satoshi Watanabe <\/p>\n\n\n\n Presentation<\/strong>: SCIS&ISIS 2026, 3 November 2026, Hachioji, Tokyo, Japan (Paper ID: 171; full paper) <\/p>\n\n\n\n Funding<\/strong>: This work was supported in part by JSPS KAKENHI Grant Numbers JP26K14983<\/a> and JP24K13350<\/a>. <\/p>\n\n\n\n
\u5e73\u5747\u30b9\u30b3\u30a2\u3068\u30d5\u30a1\u30b8\u30a3\u30a8\u30f3\u30c8\u30ed\u30d4\u30fc\u3067\u306f\u3001\u8a2d\u5b9a\u306e\u5f71\u97ff\u304c\u7570\u306a\u308b\u00a0<\/strong><\/p>\n\n\n\n
\u4e00\u65b9\u3001\u30b9\u30b3\u30a2\u304c\u300c\u4f4e\u30fb\u4e2d\u30fb\u9ad8\u300d\u306b\u3069\u306e\u7a0b\u5ea6\u91cd\u306a\u3063\u3066\u5f53\u3066\u306f\u307e\u308b\u304b\u3092\u6570\u5024\u5316\u3059\u308b\u300c\u30d5\u30a1\u30b8\u30a3\u30a8\u30f3\u30c8\u30ed\u30d4\u30fc\u300d\u3067\u306f\u3001\u9055\u3044\u306e\u5185\u8a33\u306b\u5360\u3081\u308b\u4e8c\u3064\u306e\u8a2d\u5b9a\u306e\u7d44\u5408\u305b\u306e\u5272\u5408\u304c\u3001\u5e73\u5747\u30b9\u30b3\u30a2\u306e\u5834\u5408\u3088\u308a\u5927\u304d\u304f\u3001\u7279\u306b\u6012\u308a\u3068\u9a5a\u304d\u3067\u76ee\u7acb\u3061\u307e\u3057\u305f\u3002\u00a0
\u3053\u308c\u3089\u306f\u3001\u4e8c\u3064\u306e\u8a2d\u5b9a\u3092\u72ec\u7acb\u306b\u5909\u3048\u3001\u5e73\u5747\u30b9\u30b3\u30a2\u3068\u6d3e\u751f\u6307\u6a19\u3092\u5206\u3051\u3066\u8aad\u3080\u91cd\u8981\u6027\u3092\u793a\u3059\u77e5\u898b\u3067\u3059\u3002\u305f\u3060\u3057\u3001\u3053\u306e\u6307\u6a19\u306fAI\u306e\u4e3b\u89b3\u7684\u306a\u611f\u60c5\u3084\u4eba\u9593\u306e\u5fc3\u7406\u3092\u76f4\u63a5\u6e2c\u308b\u3082\u306e\u3067\u306f\u306a\u304f\u3001\u7d50\u8ad6\u306f\u4eca\u56de\u9078\u3093\u3060\u30e2\u30c7\u30eb\u3068\u4f5c\u54c1\u306e\u7bc4\u56f2\u306b\u9650\u3089\u308c\u307e\u3059\u3002\u00a0<\/p>\n\n\n\n![\"\"](\"https:\/\/www.shimonoseki-cu.ac.jp\/sdgs\/wp-content\/uploads\/sites\/9\/2026\/11\/image.png\")
<\/figure>\n\n\n\n
\u767a\u8868\u5f8c\u306b\u306f\u3001\u4eba\u9593\u306e\u8a55\u4fa1\u3068\u306e\u6bd4\u8f03\u3001\u30ed\u30dc\u30c3\u30c8\u5bfe\u8a71\u3078\u306e\u5fdc\u7528\u3001\u7279\u5b9a\u306e\u611f\u60c5\u306b\u5bfe\u3059\u308bAI\u306e\u53cd\u5fdc\u306b\u3064\u3044\u3066\u8cea\u7591\u304c\u4ea4\u308f\u3055\u308c\u307e\u3057\u305f\u3002\u4eba\u9593\u306e\u8a55\u4fa1\u3068\u306e\u5bfe\u5fdc\u3084\u3001\u4ed6\u306e\u4f5c\u54c1\u30fb\u8a00\u8a9e\u306b\u3082\u540c\u69d8\u306e\u77e5\u898b\u304c\u5f53\u3066\u306f\u307e\u308b\u304b\u306f\u3001\u4eca\u5f8c\u306e\u691c\u8a3c\u8ab2\u984c\u3067\u3059\u3002\u00a0
\u540c\u4f1a\u8b70\u306b\u306f\u3001\u767d\u6ff5\u6559\u6388\u306e\u307b\u304b\u306b\u3082\u672c\u5b66\u30c7\u30fc\u30bf\u30b5\u30a4\u30a8\u30f3\u30b9\u5b66\u90e8\u304b\u30893\u540d\u306e\u6559\u54e1\u304c\u53c2\u52a0\u3057\u3001\u305d\u308c\u305e\u308c\u7814\u7a76\u6210\u679c\u3092\u767a\u8868\u3057\u307e\u3057\u305f\u3002\u00a0<\/p>\n\n\n\n
\u5927\u4f1a\u516c\u5f0f\u30b5\u30a4\u30c8: SCIS&ISIS 2026<\/a> <\/p>\n\n\n\n
Under the conditions studied, variation in mean scores across settings was primarily associated with persona. Variation associated with temperature alone was relatively small.\u00a0<\/p>\n\n\n\n![\"\"](\"https:\/\/www.shimonoseki-cu.ac.jp\/sdgs\/wp-content\/uploads\/sites\/9\/2026\/11\/image-1.png\")
<\/figure>\n\n\n\n