@Before
public void setUp() throws Exception {
classUnderTest = new BM25();
engine = IndriSearchEngine.getInstance(documentIndexDir);
String[] rules = new String[] {"method:dir", "mu:1000"};
retrivedDocumentId = engine.runQuery(1, rules, "nobel prize");
query = new Query("1", "nobel prize");
windowList = getWindowSegments(retrivedDocumentId.get(0), 1000);
passLM_List = engine.calcPassagesLM(windowList, true, query);
passagesInColl = getNumberOfPassages(1000);
}
protected List<Integer> getDFpass(Query query, String windowSize) throws Exception {
List<Integer> dfList = new ArrayList<Integer>();
for (String qTerm : query.getQueryTerms()) {
int df =
engine.runPassageQuery(String.format(PASSAGE_RETRIEVAL, windowSize, windowSize, qTerm));
dfList.add(df);
}
return dfList;
}
private List<WindowBoundary> getWindowSegments(RetrivalResult res, int segSize) throws Exception {
List<WindowBoundary> windowList;
FixedWindowStrategy windowSegments = new FixedWindowStrategy(segSize);
int[] docsID = new int[1];
docsID[0] = res.getIndriDocId();
ParsedDocument pDoc;
pDoc = engine.documentPositions(docsID);
windowList =
windowSegments.getWindows(
pDoc, res.getDocumentId(), res.getIndriDocId(), -3.61494, res.getEndPosition());
return windowList;
}
