Include hyponyms in semantic search results

[aarppe]

Allow for the inclusion of hyponyms (or the full hyponymic subtree) in semantic search, by using of the RW and WN hierarchy.

I'd be inclined to make this default, so that getting the exact match in terms of class would be the option that the user could select, if they so wish.

In this way, the speech-DB could already be used for semantic exploration of vocabulary content, though we would probably want to have a distinct interface that is better adapted for representing the semantic hierarchy.

[aarppe]

@nienna73 This and #407 would be an important feature and presumably relatively straight-forward to implement, significantly improving the searchability of the vocabulary content.

[aarppe]

As an example, take the following highest-level in RW:

5 Daily life 5.1 Household equipment 5.1.1 Furniture 5.1.1.1 Table 5.1.1.2 Chair 5.1.1.3 Bed 5.1.1.4 Cabinet 5.1.2 Household decoration 5.2 Food 5.3 Clothing 5.4 Adornment 5.5 Fire 5.6 Cleaning 5.7 Sleep 5.8 Manage a house 5.9 Live, stay

If one would search for the most specific RW class 5.1.1.2 Chair, one would only get those entries that are grouped under this particular class.

If one would search with the more general class 5.1.1 Furniture, one would get not only entries classified at this particular level but also the contents of the more specific classes 5.1.1.1 Table, 5.1.1.2 Chair, 5.1.1.3 Bed, and 5.1.1.4 Cabinet.

Further up, searching with 5.1 Household equipment, one would get not only anything classified under 5.1.1 Furniture and its subclasses, but also 5.1.2 Household decoration, and its subclasses, and so on.

And finally, searching under 5 Daily life would give not only all content under 5.1 Household equipment and all its subclasses, but also anything under 5.2 Food, 5.3 Clothing, and so on up to 5.9 Live, stay, and all their subclasses.

[aarppe]

As part of this, we would want to add the entire RW hypernymy hierarchy, even when some classes have not been used.

[fbanados]

Sounds like this has been implemented, but classifications need improvement, which is a different issue.