Den 2013-04-03 09:29 skrev Alex Peshkoff såhär: > On 04/03/13 11:23, Kjell Rilbe wrote: >> Den 2013-04-03 09:09 skrev Alex Peshkoff såhär: >>> On 04/03/13 01:44, Thomas Beckmann wrote: >>>> Second: I'd like to have this like a plugin architecture, >>> The most simple requirement - just write UDF (or in FB3 external >>> function). No other plugins are needed. >> I think that Thomas meant was a phonetic matching function that can be >> configured in a way similar to collations, i.e. for this language, use >> these phonetic rules. No programming, just config. > Sorry - plugin means adding _code_, not just doing configuration.
:-) I know what he wrote. I just thought he meant config, because that made more sense to me. If phonetic rules were to be programmed in code for each language, then I see little point in building a separate plugin architecture for it, because of the already existing support for udf/external function, just as you mentioned. It's in the phonetic rules the complexity lies, not in the actual comparison of the already phonetically encoded data. So, as I see it these are the sensible options at hand: 1. Do nothing. No phonetic support at all. 2. Include a phonetic matching function that supports English only. Anyone needing a different language has to write a completely separate udf or external function. This would be a rather simple task, just code double Metaphone and include it in the standard FB releases, but adds little value for non-English data. 3. Include a phonetic matching function that uses phonetic rules from some kind of config. Possibly include a few languages by default or provide them all as separate downloads. This is more complex and would probably require quite a bit of research to define how the config is to be structured/defined, but if successful would provide better value for data in various languages. 4. Include a phonetic matching function that works well with data in any language or mixed languages. This would be even more difficult than above, but if successful would provide maximum value for all languages. However, I think this would be subject to several academic research projects, one for each language, so I consider this to be undoable. Note: I'm not quite sure if double Metaphone claims to actually support mixed-language data, or if it is intended simply to support English with some foreign/imported words and/or names subject to English pronunciation. Does anyone know? Regards, Kjell -- -------------------------------------- Kjell Rilbe DataDIA AB E-post: kj...@datadia.se Telefon: 08-761 06 55 Mobil: 0733-44 24 64 ------------------------------------------------------------------------------ Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
