- implement stemmer for strings
  - implement a simple stemmer
- build an analyzer to split the strings
  ? the analyzer uses tokens to split the stream
  ? it can throw out string parts, which have no meaning
  - builds an array of strings with number of occurences (needed for scoring)
    * the index writer will do this, as i think the analyzer just has to
      deliver the array itself
* IndexSearcher
  * build indexes for attributes
    * implement tree structures for index
      * binary tree
      * b-tree
      * word tree for strings
      * index has only the position of document in storage
  * add scoring for findings
    * could be implemented as per document score stored in the index
    * findings of string in the document in relation to findings in all documents
      * info should be accessible after building the index
* implement resultset
  * should be streamlined to gather resultset from multiple queries
  * sorts the result if needed
  * returns only the top x documents