full place names carried into searches
Comments
-
Adrian Bruce said: Sorry, just for clarity, absolutely agree that we shouldn't have "Germany" as a name pre-1871!0
-
Lundgren said: I can appreciate your frustration. Hopefully this discussion helps both of us better understand the challenges of building a system that supports the needs of a specific location as well as the entire world.
Familysearch must also support a wide range of users. We cannot target just professionals or just beginners.0 -
Adrian Bruce said: I certainly feel that I understand more about the strategy behind the search parameters generated by FS Family Tree when submitting an enquiry to Historical Record Indexes. However, I remain to be convinced that the results show any benefit to anyone given the massive volume result set generated by the current two node query in any part of the globe. Having said that, even the full place-name queries generate result sets that are impractical - I just tried this on a long-term resident of San Francisco, targetting the US records only - it reduced from 1834 records in the two node results to 609 for the full place names results.
This makes me wonder if the two node search principle actually benefits the US data or not!
My suggestion for a way forward would be for FS to do an analysis of the current situation for US profiles in FSFT (since obviously this has driven the previous coding). Does restricting the submitted queries to two place-name nodes really help even US searches? Or would there be advantages to submitting the queries using the full standardised place-name?
Not totally sure how one defines "advantages" but how to work with the volume of the result set is one aspect.
I think I remember when this restriction to two nodes came in - and I also think that there were a couple of changes taking place in searching at the same time so I'm not sure like was being compared with like.
Yes, there are things like the SSDI that appear to support the looser place-name searches. However, since the SSDI does not include birth-places (unless anyone knows better) I am not convinced that SSDI searching is set up correctly in the first place (see my comments further up this thread).
Yes, I acknowledge the sense of not being too precise in place names because people move. However, my impression is that the Historical Record search algorithms already add in results for places that don't match a full place name (the SF guy got given 1920 census suggestions in San Bernadino Co. and Vallejo Co., both in CA, as well as SF). So the full length nodes are not precise. Not unless you tick the Match Exactly boxes!0 -
joe martel said: However, I would not delete a person's vital, say birth conclusion, unless you have a better value. The place may have changed names over time, but if I know my ancestor was born in say, Russian, I would rather have that than empty. That gives users a place to start and refine.0
-
Cindy Hecker said: No one search query will find everything because records are all different. From experience I play with the search engine by adding details and taking them away too. I get different results and I can usually find what I want. But for example I had a person yesterday where 3 records showed him born in a different location (Iowa in a census, West virginia in a death record, and Virginia in a different census) by only having 1 birth location in the search I would not have found all 3. But just using death information helped me and using his child's information helped me find the census record because the child was born just a year before census so his information was correct on census and Dad's was not.
Just try and play with the search and I don't think any one search or the way the search engine works will ever find everything. The user has to be the one with the creativity and knowledge a computer doesn't have!0 -
Lundgren said: This is an important piece of the puzzle.
The search parameters being sent from tree are a starting place. I don't think there is a way to make them correct.
What is there today of England, UK needs to be improved.0 -
Adrian Bruce said: Yes - Ellis Island instructions (I think) at one point required the current country to be entered, regardless of when someone was born. Clearly they weren't genealogists!
Actually in such a case one could standardise "Russia" but enter a display name of "Later Russia" as a prompt. (Decorating place-names like that is one excellent facility in FS).0 -
Adrian Bruce said: "The search parameters being sent from tree are a starting place." Totally agree.0
-
Adrian Bruce said: In fact - that's a thought. Could the "search parameters being sent from the tree" be parameterised themselves? Each patron could have an "Advanced Search Options" and one parameter would be how many nodes of the place-names to be sent through in any search of history records from the tree -
- two (as now);
- three (why not?);
- full (the full standardised place-names)
Then all us UK (ish) genealogists could set our parameters to "full". No more complicated than the current filtering mechanisms....0 -
Donald Allen Soper said: Don,t think so, that would require a programming change0
This discussion has been closed.