Akinori Abe


LSE (Language Sense processing Engineering)

In these ten or twenty years, computers have been able to deal with human language (natural language). Researches computationally dealing with natural language are important to achieve multi-lingual machine translation, automatic discourse understanding, and automatic discourse generation etc. However, in fact, such researches seem to focus on only mechanical or structural aspects of human language by means of the methods based on grammatical theories. This is because if we deal with language by using computers, it is quite natural to deal with language by logical ways. Accordingly, it will be slightly unique to deal with ``affective'' or ``emotional'' aspects that will be naturally possessed by human language in a computational language processing.
Therefore, we proposed Language Sense processing Engineering (LSE) that deals with affective or emotional aspects of language (in computers). Accordingly, LSE is interdisciplinary research that studies language from computational, affective, psychological, artistic, sociological, socio-lingual, philosophical, cognitive viewpoints. Some of the possible applications of LSE are (computational) generation of literature or poems, (computational) word plays, and (computational) generation of artistic things.

Output from B-class Machine

Definition of Language Sense

The keyword for LSE is ``language sense''. We hereby use the word language sense to show affective, emotional or psychological aspects of language. In fact, language is used as a tool for communication. Accordingly, language should correctly transfer speaker or writer's intention to listeners or readers. However, sometimes we add some effects or additions to language or intentionally remove a certain information from language. They are sometimes humour, joke, esprit or pun. These types of phenomenon cannot be logically explained. By language sense, we want to express a type of non-logical feeling in language. That is, language sense is the affective or emotional aspects of language.
Therefore, some of the targets of language sense are to deal with or to automatically generate poem, literature, humour, joke, esprit, conversation, etc.

Computational Generation of Affective Phrase

Recently, words instead of multimedia have been thought of as an important factor in communication. This is because words are cool media that has less visual information, therefore it is possible to add flexible affective or emotional information to words. Accordingly, it is important to deal with the affective or emotional aspects of language. Language Sense processing Engineering (LSE) is such a research that deals with language from the affective or emotional aspects. We showed a semi-automatic generation of catching copies as one of applications of LSE. It utilizes one of the feature of Chinese characters that have both sound and meaning in one character. In fact, conceptual base that can show associated words can be applied to it.

Select ``travel (tabi)" as a keyword. In addition, we also add viewpoint as ``waterscape (Freely selected. No theoretical reason.)''. Then, the following patterns can be generated.

Pattern 1

Pattern 2

References:

  1. Akinori Abe: Computational Generation of Affective Phrase, Proc. of SCI2002, Vol. XVI, pp. 261--266 (2002)
  2. Akinori Abe: Language Sense, Proc. of PRICAI2004 Workshop on Language Sense on Computer, pp. 1--9 (2004)
  3. Akinori Abe and Moe Hayashi: On communication via bots for the computational therapy, Proc. of Int'l Workshop on Language Sense on Computer in IJCAI2016, pp. 61--66 (2016)
  4. Hayashi Y. and Abe A.: Conversation Method Compared with Reminiscence, Proc. of Int'l Workshop on Language Sense on Computer in IJCAI2016, pp. 25--28 (2016)
  5. Akinori Abe and Moe Hayashi: On communication assistance via bots ---towards IMDJ, Proc. KES2016, Procedia Computer Science, pp. 1657--1665 (2016)