Detail

Semantic networks and spreading activation method for Question Answering

Question answering often leads to multiple possible answers that require some ranking criteria.
Semantic networks are associative knowledge bases where candidate answers can be represented, amongst other known information.
The spreading activation method can identify semantic affinity between answers along the semantic network.
This similarity can be seen as mutual reinforcement, and thus such answers are promoted by a weight increase.
The process seeks a more plausible sort order in the candidate answer list.