How to, Searching Tips for offline knowledge

How to, Searching Tips for offline knowledge http://libraries.ge.com/LibrariesWiki/3620049101/Searching%20Tips Offline knowledge uses Apache Lucene search engine internally, with the Snowball Stemmer algorithm. You can find detailed description of the Lucene query syntax here and here. By default Offline Knowledge requires all search words to appear in the document (using boolean AND), you can change that by specifying the boolean operators and grouping using parentheses. By default each search term should exist in the document as a whole or as English word or a word with matching stem. You can indicate that you may want to search for a partial word using the asterisk ('*') mark. Search example: "Service Note" (4317* OR 4315*) The above search query will return all articles with the stems of "Service Note" together (a.k.a. it can find also "servicing notes") and a token that either starts with 4317 or 4315. Note: that asterisk mark will work only if there are less then 1024 different tokens that match that sub-term. If there are more you will get "Search Error" Note: that since we are using English stem there is no way of searching for exact match, i.e. the search engine can not distinguish between the word 'search' and 'searching' For something to be found it have to be indexed, match your query, and have high enough score to be seen. What is indexed (each bullet is a different Lucene document field)? · KO ID · KO Name · Concept name, and Concept short description (from all concepts in KO) · Concept synonyms and Concept long description (from all concepts in KO) · Only words that contains alphanumeric characters and underline. What is NOT indexed (these are just example to iterate what is not indexed and you probably do not want to add in your search query)? · Concept ID · Attachment name · Attachment content · KA Name · Words that contains non alpha number characters (i.e. you can not find 'Display:0.0' for the search engine this looks like 3 different words "Display", "0" and "0") In addition the KO weight (usually an matches the number of experiences that KO have) is used to boost the weight of that document. The document score is taking into account the following: · How many times the STEM was found in the document · The length of the field where the document was found (i.e. if it is in the KO name it will probably will have more weight then in the concept long description) · The boost of the document (i.e. the KO weight) · If there are multiple terms in the query, then each term may have its own weight according to the number of documents this term belongs to (the more unique the term is the higher its score is). For more information about scoring refer to http://lucene.apache.org/java/2_4_0/scoring.html