We are looking to build 'fuzzy search' into a website. The site is built using PHP/MySQL - but we have no restrictions over which version to use.
It should be capable of the following:
- perform lookups using a search term that is passed in via our input form
- look through specified fields in a content table: different fields in the table could have a different weighting
- be able to handle misspellings, or stem differences
- be able to handle synonyms (i.e. a search for 'blue jacket' would pick up 'blue coat')
- it must perform well, and be scalable
Our initial thought for how this could work is that it would be based on the following:
- MySQL full-text searching
- PHP soundex / metaphone / levenshtein for the misspellings and stem differences
- Something like Wordnet for the synonym checking
To ensure performance is not compromised it may be necessary to do whatever we can not in realtime - e.g. generate soundex values and a synonym list for content items when they are created or when an indexing funciton is executed, rather than when we are searching.
If there is existing third party software that would help fulfil these requirements then we would be glad to consider it.
In order for this project to be deemed a success we must be able (with your help) to integrate this search into our website, and see evidence of a wide range of test input searches producing the expected results.
We are based in Sheffield, UK.
A few questions that will help us make our selection:
- Can I call 2 or 3 of your past clients for reference?
- Please provide examples of related projects
- Can you provide CV/Resume of person(s) who will actually work on this?
35 freelancers are bidding on average $3524 for this job
We are suitable for this project as we have already done same type of project...we can give you the exact solutions you are asking...Please check your PMB for more details..... Thanks
Confirm at our confidence to provide the quality work within the certain time limit. I have gone through your questions lastly. I agree to all of your questions.