Computational approach for finding a quantitative law on human word recognition

• Project/Area Number: 26540113
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: Tokyo Institute of Technology
• Principal Investigator: Takamura Hiroya 東京工業大学, 科学技術創成研究院, 准教授 (80361773)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Project Status: Completed (Fiscal Year 2016)
• Budget Amount: ¥3,640,000 (Direct Cost: ¥2,800,000、Indirect Cost: ¥840,000); Fiscal Year 2015: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000); Fiscal Year 2014: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
• Keywords: 単語認識 / 読み時間 / 自然言語処理

Outline of Final Research Achievements

We wrote computer programs : (i) a program measuring the similarity between words such as edit distance, cosine similarity of character n-gram vectors, and string kernel, (ii) a program measuring the information amount that a word has given its preceding context using the language model trained on news paper articles, (iii) a program that is used in the experiment to measure the reading time of human subjects. With an ethical approval, we conducted experiments to measure reading time, where the spellings of some words are intentionally changed. From the experimental data, we derived the statistical trend among reading time, spelling, and the information amount of a word given its preceding context.