On 26 Jun 2009, at 12:25pm, Alberto Simões wrote: > one adition, one remotion or one substitution
I am always amazed at how well people use English. For your word 'remotion' you probably mean 'removal' or 'omission'. You have joined the two possibilities together ! > Then, the script constructs an SQL query: > > SELECT DISTINCT(word) FROM dict WHERE word = "ar" OR word = "ca" OR > word LIKE "_car" OR word LIKE "c_r" OR word = "cr" OR word LIKE "_ar" > OR word LIKE "ca_r" OR word LIKE "c_ar" OR word LIKE "ca_" OR word > LIKE "car_"; > > And this SQL quer works... but not as quickly as I need (specially > because the speed is proportional to the word size). You could write a program to prepare another table in the same database with your near-misses in. In other words, to take each word in the dictionary (like 'car') and put entries in this other table for each near miss you wish to accept: nearMiss realWord -------- -------- car car ca car cr car ar car ca_ car c_r car _ar car _car car c_ar car ca_r car car_ car cat cat ca cat ct cat at cat ca_ cat c_t cat _at cat _cat cat c_at cat ca_t cat cat_ cat Then, in your search phase you just consult the near-miss table SELECT realWord FROM nearMisses WHERE [whatever] LIKE nearMisses.nearMiss; and find all the applicable entries: a single lookup against one index should be extremely fast. Look up the word 'ca' and you get the both 'car' and 'cat' realWords. You could even include a JOIN to find the entries in your dict table too. It should be easy to write software which goes through every permutation of missing letter, extra letter, etc.. It will lead to one very big table, but it will give you instant lookup. You can shrink the table by using the LIKE operator both ways around, at the penalty of doubling the time taken. The choice of whether to bother consulting the nearMiss table if the user typed a real word to start with is up to you. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
