I have a question regarding the matching. The officer search always returns a very large and fuzzy set of results. Even for fairly unique director names, you will retrieve 10,000+ results even though only the first one is an exact name match. Is there a way to specify the matching level and only return exact or full name matches? Is this a planned functionality? We would preferably want to hide the partial name matches.
I noticed that the web interface suffers from the same fuzzy matching.
Many thanks!
1 Like
I am having the exact same issue. I used my own name as a reference as I know I have 3 directorships. However, using just my surname I got 6000+ results. Fair enough I thought, there may be a lot of people with the same surname. So I added my firstname as well to narrow it down and ended up with 8000+ results.
It appears that the search will find any individual word in the search string anywhere in the title or snippet of the CH records, so the more you put in, the more matches you get.
This seems a bit mad to me. I would expect more search criteria to reduce the returned results, not increase it.
Is there any way we can make the search only find records that contain ALL of the keywords we put into the search criteria ?
Regards
1 Like
Both,
The search functionality provided is based on providing the most relevant matches first based on the search criteria provided. Therefore the number of results will increase with additional search parameters provided, however the most appropriate result will appear first. So for example, if I search for Fairhurst I would expect all officers with the surname of Fairhurst to be returned, and if I then search for Mark Fairhurst, I would expect Mark Fairhurst to be returned first followed by all Fairhurst’s, then all Marks.
We are currently improving the search algorithms to ensure the relevance described above is achieved consistently, and have deployed new algorithms to the disqualified directors search, which we will be rolling out to the other search indices in due course. Please give this a go and feedback your thoughts.
To be clear we are not providing search which allows for exact match returning only. The strategy is the search result order should be consistent and predictable allowing you to simply cut off the results to your specific requirements.
Thanks
@mfairhurst
I appreciate some of the reasoning around providing more data than less as long as it’s mitigated with predictability. I do have a question related to this:
allowing you to simply cut off the results to your specific requirements.
Are there guidelines on how this would be achieved? i.e. Do we just simply scan the result sets until we meet a non-exact match? Because I can see a lot of problems with that, including that a user might see a count of 6000 but a result set of 125. As well as the fact that sometimes we’d need to fetch multiple pages in order to get to this “cut off point” which incurs more API calls (costs) on the consumer side. None of which are insurmountable, all of them unpleasant though.
The link in your reply doesn’t work, is it potentially this one: Companies House Public Data API: Search disqualified officers ?
This is a common issue and we’d obviously want to be able to isolate these matches somehow, but it’s unclear what the guidelines are besides trawling through the pages.
Thank you.