Hi,
the problem I'm facing is, that my customer is complaining about the
quality of "did you mean" suggestions.
He is typing 'sammsuh' and expects to see a suggestion for 'samsung'. He is
arguing, that google is able to suggest samsung
(https://www.google.de/webhp#q=sammsuh).
I think the issue is, that the phrase suggester only allows for edit
distances <= 2, whereas the edit distance of 'sammsuh' and 'samsung' is 3.
So my question is whether it would be possible to allow for higher edit
distances.
He is arguing, that google is able to suggest samsung
Not a fair comparison. Google suggestions rely heavily on recycling user
behaviour (people who searched for X typically clicked/re-searched on Y)
rather than simple string comparisons of your content.
See http://searchengineland.com/how-google-instant-autocomplete-suggestions-work-62592
If you have a lot of user behaviour data that is typically a smarter way of
making suggestions.
On Thursday, September 25, 2014 11:41:19 AM UTC+1, Felix wrote:
Hi,
the problem I'm facing is, that my customer is complaining about the
quality of "did you mean" suggestions.
He is typing 'sammsuh' and expects to see a suggestion for 'samsung'. He
is arguing, that google is able to suggest samsung ( Google).
I think the issue is, that the phrase suggester only allows for edit
distances <= 2, whereas the edit distance of 'sammsuh' and 'samsung' is 3.
So my question is whether it would be possible to allow for higher edit
distances.
He is arguing, that google is able to suggest samsung
Not a fair comparison. Google suggestions rely heavily on recycling user
behaviour (people who searched for X typically clicked/re-searched on Y)
rather than simple string comparisons of your content. See http://searchengineland.com/how-google-instant-autocomplete-suggestions-work-62592
If you have a lot of user behaviour data that is typically a smarter way
of making suggestions.
On Thursday, September 25, 2014 11:41:19 AM UTC+1, Felix wrote:
Hi,
the problem I'm facing is, that my customer is complaining about the
quality of "did you mean" suggestions.
He is typing 'sammsuh' and expects to see a suggestion for 'samsung'. He
is arguing, that google is able to suggest samsung ( Google).
I think the issue is, that the phrase suggester only allows for edit
distances <= 2, whereas the edit distance of 'sammsuh' and 'samsung' is 3.
So my question is whether it would be possible to allow for higher edit
distances.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.