The ability to use logical deduction is important for various intelligent systems. Specifically, the ability to apply logic inference rules to get correct arguments. This is necessary in decision support systems, various expert systems etc. Sometimes it is not useful to apply existing systems of inference rules. Non-traditional rules from experts may give much better results in some cases, especially if the expert is using the “reasoning technology” is based on experience from the specific field. However, these experts may not be able to identify and formally describe their deduction mechanisms.
The authors look at mining inference rules from natural language texts. They argue that it is possible to use the DST dialogue system to search for logic formulas and inference rules from natural language texts. In order to explain the extent and reliability of this approach, they use the RS-meta-procedure that was created for system mining.
Published in: The 3rd International Multi-Conference on Engeneering and Technological Innovation.
Lorents, P. & Matsak, E. (2010). System mining inference rules from natural language texts. The 3rd International Multi-Conference on Engeneering and Technological Innovation. International Institute of Informatics and Systemics, 2010, pp. 309 – 314.