Improvements to English
This release brings a number of improvements to English:
Better negation: we've added several English negations (such as "nothing" and "unlike"), as well as adding "only" to the list of words that stop negations (so that "not only a doctor" will be the concept "doctor", not "not doctor").
Better stopwords: some words will start being stopwords, such as "like" when it's used as a preposition (but not as a verb!), and "us". Others will start being stopwords because we've added them to our list, such as phone-transcript words like "yep" and "um", and a handful of other filler words like "other" and "today".
Better stemming: more sophisticated stemming especially for multiword prepositions like "according to" and "in front of", which means projects won't be cluttered with terms like "accord" and "front" when they're not meaningful.
This impacts all new projects and any existing projects that are “upgraded” by Luminoso. Note: to upgrade your existing projects and take advantage of these new changes, please reach out to Customer Success at email@example.com.
Luminoso Analytics and Luminoso Compass Differences in how stopwords are treated as collocations Starting this release the stopwords present in phrases are ignored, which leads to more unified understanding of similar fragments.
Example: "consider problems", "consider the problems", and "consider all the problems" will be treated as the same.
Changes to what is included in "create" and "write" permissions
The meanings of the permissions have changed somewhat, such that "create" permission is required for actions that count toward the account's number of documents uploaded, and not for actions that don't.
The primary effects of this change are:
Users with "write" permission are now allowed to copy existing projects, they no longer need "create" permission.
- Users with "write" permission are no longer allowed to add data to existing projects, this now requires "create" permission.
Added ability to add and modify users with "create" permissions
Previously users with “account_manage” permission could only grant other users “read" and "write” access. Now they can also grant “create” permissions, as well as delete users who have “create” permissions. Luminoso Analytics API
Changes to JSON return format for GET term document counts
The return format of the GET /api/v4/projects/<account>/<project_id>/terms/doc_counts/ endpoint in JSON mode has been changed.
Previously this endpoint returned a JSON dictionary object from term texts to count data. We have changed the return format in order to support multiple terms having the same text -- typically, related word forms with the same spelling, such as “training” in “employee training” (the noun “training”) vs. “training an employee” (the verb “train”). The endpoint now returns an array of JSON dictionary objects, each with the term text as one of the elements, and with the same count data as before.