FAQ
1. What is the goal of the game?
2. I added a word to the graph, but it didn't form any links. What gives?!
3. How does Linxicon determine how similar words are to each other?
Linxicon compares words using three signals: word vectors from ConceptNet Numberbatch, which represent patterns of meaning and association; dictionary relationships between meanings in WordNet; and shared WordNet topics, which count only when the word vectors also show some similarity.
The final score is the strongest of these three signals, rather than their average. Some dictionary relationships receive fixed scores. The percentage is a game relationship score, not a probability or a measure of how interchangeable two words are. The same pair of words receives the same score regardless of who is playing.
4. Why do some connections seem unusual?
A word can have several meanings, and Linxicon considers multiple dictionary senses without knowing which one you intended. An unfamiliar or figurative meaning can create a strong link even when the meaning you had in mind seems unrelated. Words can also be related without being synonyms.
For example, “still” can mean “not moving,” but it is also the name of equipment used for distilling. That second meaning explains a connection to “distillery,” even if you were thinking of standing still.
The score reflects relationships in the source data, so it may not match the association you had in mind. If a link seems surprising or incorrect, please let me know which words were involved so it can be investigated.
5. Is it possible to lose?
6. Why are some words not allowed?
7. This game is hard! Can I get a clue?
8. Does Linxicon have a privacy policy?