| 研究課題/領域番号 |
16K16089
|
|---|
| 研究機関 | 大阪府立大学 |
|---|
研究代表者 |
オジュロ オリビエ 大阪府立大学, 工学(系)研究科(研究院), 特認助教 (10772436)
|
|---|
| 研究期間 (年度) |
2016-04-01 – 2018-03-31
|
| キーワード | eye tracking / reading analysis / reading understanding / adaptive document |
|---|
| 研究実績の概要 |
I extended the research about the reading life-log, which consists in recording the reading behavior by using an eye tracker. I worked in some solutions to improve the accuracy of the eye tracker with a post-processing algorithm.
Then I developed a new algorithm for analyzing the English skill of non-native speaker. Our system is able to predict the TOEIC score quite accurately. We also developed a method to predict the which words the reader feel difficult. Then I started new experiments to analyze the reading comprehension and text difficulty of Japanese texts.
With a researcher from Bordeaux University we are building a software to estimate the vocabulary of a reader and then to recommend reading a new text by comparing the reader's vocabulary and the document words.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
Some parts of the analysis such as predicting the English or Japanese understanding worked quite well. The prediction of the TOEIC is very accurate, we are even able to predict which specific words the reader feel difficult.
However the analysis of the emotions of the reader seems harder than I expected even if some state of the art methods looks encouraging.
I am also thinking about the different way to make a smart and adaptive document. The first idea was to change the content but another idea could be to change the order of the lesson chapters depending on the reader, in order to help him to understand in a better way.
|
| 今後の研究の推進方策 |
The analysis of the reader understanding and language skill is working quite well with a stationary eye tracker. We plan to make some experiments by using JINS MEME Electro-oculography glasses to see to which extent, a simpler device could be use for the same analysis. We know that we can predict in the same way the number of read words but no research have been to to use it for analyzing the reader understanding.
The second part of the future work is to develop the actuation part of the smart documents. We successfully developed algorithm to analyze the reader behavior, now we need to see how this information can be used to change the document and make them fit the reader.
|
