2013 Fiscal Year Final Research Report
Algorithm for an automatic moral judgement of human behavior by using web-mining
| Project/Area Number |
24700253
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Cognitive science
|
|---|
| Research Institution | Hokkaido University |
|---|
Principal Investigator |
RZEPKA Rafal 北海道大学, 情報科学研究科, 助教 (80396316)
|
|---|
| Project Period (FY) |
2012-04-01 – 2014-03-31
|
| Keywords | 知識獲得 / マシンエシックス / 人工汎用知能 / 自然言語処理 / コモンセンス |
|---|
| Research Abstract |
The idea proposed in this study was an ethical reasoning algorithm of human behavior. The behavior of the system, is determined by a collection possible "ethical common sense" from Internet resources rather than by the instructions of the programmer as in conventional systems. Many Internet users have accumulated their "folk wisdom" which is the base for the proposed system. Feedback to the system is minimized by utilizing "automatic emotional evaluation of an act's results" algorithm and "automatic social evaluation of an act's results" to extract and the resulting reaction patterns of human behaviors. This helps the system to determine whether the given act is regarded "good" or "bad" by humans (no limit to the topic in the input). For deeper situation understanding five senses simulation and instincts lexicon was added and methods for calculating vectors of Bentham's Felicific Calculus were proposed. Precision achieved was 70-75% on average for each recognition step.
|
