A function that brings up words that are similar in usage from a certain word

< Assumed procedure>
(1) Prepare the original word (seed) (including nouns, verbs, adjectives, and sentence expressions if possible)
(2) Estimate and output words that are used in the same way as Seed from the input sentences.
(3) Check the words output in (2), and specify the words you expect and the words you don't.
The assumed words will be used as seeds from the next process, and the other words will not be output from now on.
(4) By outputting a list of words that are Seeds, a list of similar expressions can be obtained.
(5) By preparing seeds for multiple types of categories, you can obtain results that are summarized without having to set night batches one by one.

The important thing is that in (3), since it is assumed that the process will be repeated, the unnecessary words will not be added to the candidates after that, and the words that have been accepted as OK will be set as seeds and the next process will be executed. 
This is the part where automatic processing such as the night batch of (5) is provided.

<Usage example>
If you want to collect and categorize the types of positive and negative expressions, set positive and negative expressions.
(1) Prepare expressions such as “like” and “delicious”.
(2) Expressions such as "preferable", "addictive", and "enjoyable" are output.
(3) When "enjoyable" is assumed to be otherwise, words such as "enjoyable" are not displayed as candidates thereafter.
"Preferable" and "addictive" are used as seeds.
(4) As a result, expressions such as "like", "favorable", "delicious", and "addictive" can be grouped into categories.