Calculate ratios, such as «find a word D related to the word C in the same way as the word A is related to the word B». An example is given in the placeholder: which word is in the same relation to the word «father» as «daughter» is to «mother»? The answer is «son». More on this...
You can also think of this operation as removing the «mother» component from the semantics of «daughter» and augmenting it with «father» component. The model inferences that it should change the gender. Thus, it outputs «son» as an answer.
Choose the model:
If you feel confident with algebraic operations on vectors, you can try something more sophisticated than simple analogical inference.
Enter not more than 10 space-separated words into positive and negative forms. WebVectors will sum up vectors for the positive words and subtract vectors from the negative ones. Then it will output the word closest to the resulting vector. If you leave negative form empty, WebVectors will simply find the center of word cluster formed by your positive words.
Language Technology Group
Nordic Language Processing Laboratory