Reputation: 35
I try to list the (physical) characteristics relevant to a given noun, i.e., all the words (adjectives and nouns) that could describe this noun. For example:
road
... should return:
black, flat, tar, porous, long...
I've tried several unsatisfying approaches. For example, dictionaries should provide such words, but their definitions are too strict and too short. I also tried semantic networks such as ConceptNet (see this link). But to the best of my knowledge, the available resources (for my application) are scarce. I was never able to predict that a road is (usually) black.
Do you have any better idea/approach?
If this question is too naive, please excuse my poor knowledge of the field.
Note: In my understanding, the problem is difficult because it deals with the "common sense": I get less Google results for "black road" (1.8m results) than "white road" (2.4m results) because nobody talks about the black roads (too obvious!). As a result, the training of the machine is biased.
Upvotes: 1
Views: 46