MétaCan
Menu
Back to cohort
Record W7134606798

深い意味処理のための統合日本語コーパスの開発と言語研究

2017· other· en· W7134606798 on OpenAlexaboutno aff
キョウコ オハラ, Kyoko Ohara, 京子 小原

Bibliographic record

VenueKeiO Associated Repository of Academic Resources (Keio University) · 2017
Typeother
Languageen
Field
Topic
Canadian institutionsnot available
Fundersnot available
KeywordsFrameNetSemantic role labelingAnnotationParsingNatural languageResource (disambiguation)Frame (networking)Principal (computer security)
DOInot available

Abstract

fetched live from OpenAlex

本研究の究極的な目的は, 研究代表者が構築した認知言語学に基づく詳細な意味情報を付与した言語資源「日本語フレームネット」(JFN)を, 言語研究のみならず自然言語処理にとってもこれまで以上に有益な言語資源とする方策を検討することである。そのためには, JFNの規模を大きくさせることと, JFNの持つ豊富な意味情報に加え統語情報をも含める, という2つの方向性が考えられる。本研究では特に, 国立国語研究所で構築中のNINJAL Parsed Corpus of Modern Japanese (NPCMJ)とリンクさせることを検討した。NPCMJは文の統語情報のみで意味情報は含まないが規模は大きいのに対JFNは詳細な意味情報を含むが規模が小さく, 両者を統合することで深い意味処理を行う自然言語処理アプリケーションのための言語資源とすることができると考えたからである。 当初計画したJFNとNPCMJのリンク方法は2つあった。一つ目は, JFNで意味情報付与の際の入力データとしてNPCMJの統語情報付きデータを用いることである。二つ目は, JFNとNPCMJを統合させ, 認知科学的にも妥当な詳細な意味情報と文の統語情報の両方を含んだコーパスとすることである。前者に関しては, ツールを開発した。後者については実現までの目処がたった。 本研究の経過・成果は, 国際認知言語学会, マルチリンガルフレームネット会議, 言語処理学会第24回年次大会にて発表した。 本研究によりNPCMJとJFNの統合は実現可能であることが明らかとなった。今後はJFNを深い意味処理を行う自然言語処理アプリケーションにとってさらに有益な資源とするため, より規模の大きく自然言語処理分野で定評のある言語資源とのリンクを検討したい。 The principal investigator (PI) of this research project has been developing Japanese FrameNet (JFN), a language resource that contains rich semantic information which is based on the theory of Frame Semantics. The ultimate goal of this research project is to enable JFN to be useful for natural language processing (NLP) applications, especially for those that involve deep semantic processing. In order to achieve this goal, there are two directions. One is to make the size of JFN larger. The other is to add syntactic information to JFN, in addition to the existing semantic information. This research project pursued the latter approach and focused on ways to link the JFN data with the NINJAL (National Institute for Japanese Language and Linguistics) Parsed Corpus of Modern Japanese (NPCMJ). NPCMJ does not contain semantic information at all but includes detailed syntactic information of sentences and its size is larger than that of JFN. Therefore, it was expected that the combined language resource would be useful for various NLP applications. There are two ways to link the JFN data with that of NPCMJ. One is to use the NPCMJ data as input to the JFN annotation (tagging) process. The other is to integrate JFN and NPCMJ to build a combined corpus. As for the former, a tool was implemented to allow the NPCMJ data as input to the JFN annotation process. As for the latter, how to combine the two resources was investigated. The accomplishments of the project were presented at ICLC14 (Tartu, Estonia ; July 2017), the Multilingual FrameNet meeting (Vancouver, Canada ; Aug. 2017), and NLP2018 (Okayama, Mar. 2018). Through this research project, we were able to confirm that NPCMJ and JFN can be combined. Future plans include attempting to link JFN with a language resource which is more suitable for semantic-processing NLP applications and which is larger than NPCMJ.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.001
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow), Research integrity
Consensus categoriesResearch integrity
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Not applicable · Consensus signal: Not applicable
GenreCandidate signal: Other · Consensus signal: Other
Teacher disagreement score0.013
Threshold uncertainty score0.999

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.001
Meta-epidemiology (narrow)0.0010.001
Meta-epidemiology (broad)0.0020.001
Bibliometrics0.0040.001
Science and technology studies0.0010.002
Scholarly communication0.0000.000
Open science0.0050.001
Research integrity0.0050.003
Insufficient payload (model declined to judge)0.0000.001

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.012
GPT teacher head0.221
Teacher spread0.208 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; both teacher heads agree on what is shown here.

Study designNot applicable
Domainnot available
GenreOther

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2017
Admission routes1
Has abstractyes

Explore more

Same venueKeiO Associated Repository of Academic Resources (Keio University)French-language works237,207