Optimizing inserts in fragment Table - Problem

Hello, I am experimenting with a JDBC-driven Fulltext-Indexing engine and now I have the following Problem: there is a table with my search fragments, like this: nr fragment 1 ABCD 2 BCD 3 EFGH 4 FGH Both columns are indexed. The principle Search uses a select nr from fragments where fragment like 'ABC%' So there is allway the beginning of a fragment matching and the index can be used for speedup. Now I want to insert new Fragments in the index. So I create a insert table (let's call it it) consisting some new fragments. Maybe this looks like this: ABCDE BC X F Since I know that I only look for the beginning of a fragment, I do not need to insert "BC" because a "LIKE 'BC%' would match "BCD" as well - and that is enough for me. Even, if I change the "ABCD"-Fragment to "ABCDE", I do no longer need the Fragment "ABCD" by analog reason. So the correct result of the main fragmen table shall be: 1 ABCDE 2 BCD 3 EFGH 4 FGH 5 X Now my question is: how to implement this in SQL? I use Informix Dynamic Server 7.23 and Informix JDBC 1.5-JC1 with JDK 1.1. I even have a Syntax Problem to create a LIKE-Clause which checks if the value in one column of a table matches the beginning of the value in another column, e.g. something like: select nr from fragments,it where fragments.fragment LIKE it.fragment ||'%' This syntax does not work - the % sign is not interpreted as a wildcard as I would like. Any hints?