| Item type |
SIG Technical Reports(1) |
| 公開日 |
2025-12-08 |
| タイトル |
|
|
言語 |
ja |
|
タイトル |
プロンプト最適化を用いた検索クエリと検索目標の関係性アノテーション |
| タイトル |
|
|
言語 |
en |
|
タイトル |
LLM-based Annotation of the Relationship Between Search Queries and Search Goals Using Prompt Optimization |
| 言語 |
|
|
言語 |
eng |
| キーワード |
|
|
主題Scheme |
Other |
|
主題 |
情報抽出 |
| 資源タイプ |
|
|
資源タイプ識別子 |
http://purl.org/coar/resource_type/c_18gh |
|
資源タイプ |
technical report |
| 著者所属 |
|
|
|
トヨタ自動車株式会社 |
| 著者所属 |
|
|
|
トヨタ自動車株式会社 |
| 著者所属 |
|
|
|
トヨタ自動車株式会社 |
| 著者所属 |
|
|
|
トヨタ自動車株式会社 |
| 著者所属 |
|
|
|
名古屋大学 |
| 著者所属 |
|
|
|
名古屋大学 |
| 著者所属 |
|
|
|
名古屋大学 |
| 著者所属(英) |
|
|
|
en |
|
|
Toyota Motor Corporation |
| 著者所属(英) |
|
|
|
en |
|
|
Toyota Motor Corporation |
| 著者所属(英) |
|
|
|
en |
|
|
Toyota Motor Corporation |
| 著者所属(英) |
|
|
|
en |
|
|
Toyota Motor Corporation |
| 著者所属(英) |
|
|
|
en |
|
|
Nagoya Uniersity |
| 著者所属(英) |
|
|
|
en |
|
|
Nagoya Uniersity |
| 著者所属(英) |
|
|
|
en |
|
|
Nagoya Uniersity |
| 著者名 |
中西,亮輔
鈴木,結友
鷲見,優一郎
光田,英司
二宮,由樹
曽根,悠太郎
三輪,和久
|
| 論文抄録 |
|
|
内容記述タイプ |
Other |
|
内容記述 |
In recent years, large language models (LLMs) have attracted attention for their outstanding performance in many tasks such as translation and summarization. In services deployed in society, LLMs have replaced human work, and improved productivity. This paper reports the results of applying prompt optimization methods―Automatic Prompt Engineer and Automatic Prompt Optimization―to an annotation task labeling the relationship between search queries and search goals. We found that the performance of Claude 3.7 Sonnet is close to human performance (Cohen's Kappa coefficients are above 0.6) in labeling whether the words corresponding to the conceptual hierarchical relationship are included in search queries and whether the words corresponding to the attributes are as well. |
| 論文抄録(英) |
|
|
内容記述タイプ |
Other |
|
内容記述 |
In recent years, large language models (LLMs) have attracted attention for their outstanding performance in many tasks such as translation and summarization. In services deployed in society, LLMs have replaced human work, and improved productivity. This paper reports the results of applying prompt optimization methods―Automatic Prompt Engineer and Automatic Prompt Optimization―to an annotation task labeling the relationship between search queries and search goals. We found that the performance of Claude 3.7 Sonnet is close to human performance (Cohen's Kappa coefficients are above 0.6) in labeling whether the words corresponding to the conceptual hierarchical relationship are included in search queries and whether the words corresponding to the attributes are as well. |
| 書誌レコードID |
|
|
収録物識別子タイプ |
NCID |
|
収録物識別子 |
AN10115061 |
| 書誌情報 |
研究報告自然言語処理（NL）
巻 2025-NL-266,
号 26,
p. 1-6,
発行日 2025-12-08
|
| ISSN |
|
|
収録物識別子タイプ |
ISSN |
|
収録物識別子 |
2188-8779 |
| Notice |
|
|
|
SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc. |
| 出版者 |
|
|
言語 |
ja |
|
出版者 |
情報処理学会 |
IPSJ-NL25266026.pdf (924.3 KB)
