1

Topic: Search return results - puzzling

Hi

I'm still having a few issues with the search functionality:

1. When I search for Tachina fera I would expect a few results but 293 seems a tad excessive and will only confuse readers. Also, surrounding the string with quotation marks, which would usually make the routine search for that specific string, just returns the same 293 values.

2. Each individual row on the search results seems to contain information that's a bit confusing. On my sandbox I return the information in a more user-friendly way, by grouping results by the taxa they occur on - ordered to give the most likely/correct name at the top. Consider this search for "Tachina" which returns a genus, subgenus, accepted / recommended names and junior synonyms:  http://nbn-sd-dev.nhm.ac.uk/taxonbrowse … ch=tachina

Best wishes,
Chris R.

Chris Raper, Manager of the UK Species Inventory, Angela Marmont Centre for UK Biodiversity,
Natural History Museum, Cromwell Road, London, SW7 5BD.  (tel: 020 7942 5894)
also Tachinid Recording Scheme (http://tachinidae.org.uk/)

2

Re: Search return results - puzzling

Thanks for the feedback. Its been logged and we are exploring ways to improve the search functionality.
Giselle

3

Re: Search return results - puzzling

Just to follow up, David Martin has been in touch and I have suggested ways that he can improve the system. However, I am very worried that the current system will be very confusing because it appears to "fuzzy match" across all data (not just taxa) regardless of whether this is needed. It also doesn't seem to prioritise the results properly. If the user has typed something that returns no results then by all means use fuzzy matching but where there is an entirely logical species that can be returned then just return that. Equally, if the user types something with multiple return results then prioritise them so that the commonest species are listed first.

Chris Raper, Manager of the UK Species Inventory, Angela Marmont Centre for UK Biodiversity,
Natural History Museum, Cromwell Road, London, SW7 5BD.  (tel: 020 7942 5894)
also Tachinid Recording Scheme (http://tachinidae.org.uk/)

4

Re: Search return results - puzzling

The search function has been amended so that any search conducted in the Species tab will return only species results.

You can still use the search box on the front page to search the wider NBN Atlas.

In both cases the use of the * will act as a wild card to bring back differing results.

5

Re: Search return results - puzzling

Thank you - this is appreciated.

Christine

6

Re: Search return results - puzzling

Well this function is still not fixed.  A search for in Tachina fera on the main Atlas screen still asks you to chose from 5 results, and when you essentially reselect "Tachina fera" from the resulting list you still have 294 results to chose from, most of which have absolutely no relevance to the species being searched for. 

Doing the same search from the "Species" screen returns 293 results!

If the species searched for is a valid species onthe UKSI list, why on earth is the Atlas seeminging tyring to find ever iteration or combination of "Tachina" and "fera" it can.  This is very poor - it needs fixing.

7

Re: Search return results - puzzling

I agree 100% with Matt - fuzzy searching is fine but should only be attempted when no results or only poor quality results are returned using a straight text search. If a user types Tachina then trust that they know what they are doing and give them what matches in the database, ranked by the most likely/popular result. Searching for "Tachina" should NOT list junior synonyms higher than recommended names, just as searching for "Sparrow" shouldn't list rare vagrants before the House Sparrow. The UKSI contains the flags needed to do this already and it worked perfectly well on the Gateway.

Chris Raper, Manager of the UK Species Inventory, Angela Marmont Centre for UK Biodiversity,
Natural History Museum, Cromwell Road, London, SW7 5BD.  (tel: 020 7942 5894)
also Tachinid Recording Scheme (http://tachinidae.org.uk/)

8

Re: Search return results - puzzling

Thank you for your comments.

I have passed your feedback on to the developer so that they can be included in work undertaken to improve search functionality.

Giselle

9

Re: Search return results - puzzling

Hi Giselle

Is there any news on a fix for this problem? Everything seems to have gone quiet but this is a significant problem, in my books.

For instance, when you search for "Tachina magnicornis" why should a junior synonym precede a recommended name?
https://species.nbnatlas.org/search/?q= … agnicornis

Why does a search for "Sparrow" still list rare vagrants above the common House Sparrow?
https://species.nbnatlas.org/search?fq= … ;q=sparrow

This problem doesn't create an error message but it is a significant loss of credibility with users when the primary search routine gives back misleading or poor results. As I said, the NBN Gateway listed things very well and I have my own scripts that give much more logical search results than this.

Many thanks
Chris R.

Chris Raper, Manager of the UK Species Inventory, Angela Marmont Centre for UK Biodiversity,
Natural History Museum, Cromwell Road, London, SW7 5BD.  (tel: 020 7942 5894)
also Tachinid Recording Scheme (http://tachinidae.org.uk/)

10

Re: Search return results - puzzling

Hi Chris

Just following up your message re an update on search results.  Our developer has just notified us that he has deployed a boost to results based on occurrence count which now seems to be giving a better result.  Please do have a look at it and advise us of any further issues you come across.

Many thanks for your input.

Christine

11

Re: Search return results - puzzling

I have been getting some strange results, the occurrence weighting is biasing the search so that occurrence number overides the name matching.
E.g. Search for Carsia sororiata  produces the following:
Physcia caesia (Hoffm.) Fürnr. (1567)
Postia caesia (Schrad.) P. Karst. (152)
Carsia sororiata (Hübner, 1813) (330)
Rosa canina x caesia = R. x dumalis Bechst. (326)
Rosa caesia subsp. caesia Sm. (647)
Carsia sororiata subsp. anglica Prout, 1937 (42)

Not sure why caesia should come before carsia? The fuzzy matching is producing gibberish.

Ideally the search should produce Carsia sororiata (Hübner, 1813) (330) followed by Carsia sororiata subsp. anglica Prout, 1937 (42) before any of the others, particularly as my search term was Carsia sororiata.

This is not an isolated example, try searching for Caryocolum alsinella.
The top of the list is  Platytes alpinella (Hübner, [1813]) and Caryocolum alsinella is 7th in the list after other Caryocolum species.

It is really important that the search function is sorted as a matter of priority. If you want the Atlas to be accessible to everyone, the search has to be robust and produce the expected results; so that the name I type in the search box appears at the top, followed by the various synonymns, varieties etc. It should not be necessary to scroll down the list to find it.

12

Re: Search return results - puzzling

An addendum to my last post.
I use the Atlas regularly when checking records of unfamiliar species. This afternoon I have been checking a list of freshwater algae, at best they on the UK species list but generally even if they are listed there are no records in the Atlas database.
These obscure species seem to set a challenge to the search system.
The best example this afternooon  was Cylindrocystis gracilis which produced a list of 501 items. Using best match, it still had not appeared by page 10 and using species name it finally appeared on page 9.
Cylindrocystis gracilis is the correct accepted name, so why was it not top of the list?

Christine

13

Re: Search return results - puzzling

Hi Christine,

Cylindrocystis gracilis is now appearing at the top of the list...

Cylindrocystis gracilis (accepted name Cylindrocystis gracilis)
followed by Cylindrocystis gracilis I.Hirn, 1953
etc.,

but it appears that there are no records currently available on the NBN Atlas for these species.  It doesn't necessarily mean that records will automatically be on the NBN Atlas if they are in the UKSI.

Mandy

14

Re: Search return results - puzzling

I'm a little puzzled by this one as well - why are there two entries for Cylindrocystis gracilis in the search results, both of which link to the same key - and why is there two entries for Cylindrocystis obesa (as C. gracilis) but one appears at position 4 and another appears at position 10 in the list rather than next to each other as C. gracilis?

Charlie Barnes
Information Officer
Greater Lincolnshire Nature Partnership

15

Re: Search return results - puzzling

Hi,

We have been working to rectify the issues with the taxonomy and integrating the species inventory into the NBN Atlas, however, this is proving to be a more complex job than originally thought and so far we have been unable to find a solution that works satisfactorily. We are still investigating and hope to have a solution in the near future.

Thank you for your patience and support in the meantime.

Regards,
Mandy