✔ Problems with front-end search
Completed by Morgan M.
- Assigned to
-
Anil V.
- Notes
-
Further to the video below, there are three immediate problems with the front-end search and one long-term problem. The immediate problems are the following:
- All tags should not be listed as soon as the user selects the search field (suggested tags should only appear once a few characters are entered that match text in the tags)
- Searching for a term that is included in the text of a tag does not produce a tag match. I believe a tag match should be produced if any text in the tag matches the search term (However, I'm open to discussing this issue -
Ryan
Melissa
- When sorting search results by relevancy, I would expect documents with tag matches to appear first in the list.
More long-term, we should consider refining the relevancy ordering by creating an algorithm that ranks documents based on both the number of keyword matches and tag matches, and assigning score values to each type of match to determine their ranking.
Typically a keyword search would only search the actual text of the document but I can certainly understand a use case for also displaying results with tags that partially match the search.
Since in this case keyword matches and tag matches are called out separately within each search result I agree that partial tag matches should be returned in the search result. If multiple tag matches are found within one result, the tags should display as a comma separated list.
Melissa
Also, I'm I correct in stating that the link to "see all search results for "termination" no longer relevant, because the search is performing an "all search" by default:
Thanks,
Morgan
We reviewed and understood above issue. We require to change existing algorithm to implement this change and hence would require more efforts. It would be good if we can plan this for next week. Please suggest.
Morgan
This has been uploaded on p3dev. Please review and feedback.
The changes on p3dev look good. I think we still have to look more closely at the relevancy issues, and creating a more nuanced algorithm that ranks documents based on a combination of tags hits and keyword hits (it appears we have currently simply given priority to tags hits), but we can work on this at a later stage in development. Please go ahead and migrate the changes to ilg.com
Also, I noticed that the speed of the search using the example from the video has slowed down quite a bit. Is there a way to speed this up? I'm concerned that the search will slow down further when start adding more projects/documents post-prototype.
Thanks,
Morgan
Morgan