2018 Fiscal Year Final Research Report
Linguistic information extraction and digital archive on Rongorongo
Project/Area Number |
15K00472
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Library and information science/Humanistic social informatics
|
Research Institution | University of Nagasaki |
Principal Investigator |
|
Project Period (FY) |
2015-04-01 – 2019-03-31
|
Keywords | ロンゴロンゴ / 自然言語処理 / 失われた言語 |
Outline of Final Research Achievements |
The target of this study is the sequence of symbols engraved on wood artifacts. For digital archivement, an experiment to measure the 3D shape of a more clear symbol engraved on a model with SfM is conducted. In order to recognize the engraved symbols, a method using the difference in height from the local average surface of the shape is tried. As a linguistic information extraction, the problem of classifying symbols into letters is tackled. Decoding undeciphered characters is a difficult task to evaluate, because the correct answer is unknown, and we use research methods to evaluate the method itself using a known language. Therefore, we prepared the data necessary to conduct an experiment to extract linguistic information from character images of known languages. We also made a presentation in order to let computer-oriented research on lost languages be known widely.
|
Free Research Field |
計算機科学,自然言語処理
|
Academic Significance and Societal Importance of the Research Achievements |
本研究課題は、未解読言語の解読という、従来人間の考古学者が行ってきた知的活動を部分的に自動化しようとする試みである。情報機器を使うことの利点は、ネットワークを介した情報の共有や膨大な計算を遂行できることだけでなく、明確に定義された手順であれば、対象を変えても同じように遂行できることにもある。対象が変わっても問題の質が変わらないことを慎重に検討する必要があるが、従来人間の手で行われていた作業を自動化することによって、判断の揺れや恣意の混入を避けることができる。そういう方向での計算機(言い換えれば人工知能)利用を進めるきっかけになるものと考えている。
|