Changeset 1124 for trunk/gsdl/src
- Timestamp:
- 2000-04-18T16:04:29+12:00 (24 years ago)
- Location:
- trunk/gsdl/src/mgpp/text
- Files:
-
- 2 edited
Legend:
- Unmodified
- Added
- Removed
-
trunk/gsdl/src/mgpp/text/Terms.cpp
r927 r1124 44 44 stemMethod = 0; 45 45 matchDocs = 0; 46 termFreq = 0; 46 47 } 47 48 48 49 ostream &operator<< (ostream &s, const TermFreqData &t) { 49 50 s << "<" << t.tag << ">\"" << t.term << "\"stem(" 50 << t.stemMethod << ")docs(" << t.matchDocs << ")"; 51 << t.stemMethod << ")docs(" << t.matchDocs << ")" 52 << "count("<<t.termFreq<<")"; 51 53 return s; 52 54 } … … 56 58 (t1.term == t2.term) && 57 59 (t1.stemMethod == t2.stemMethod) && 58 (t1.matchDocs == t2.matchDocs)); 60 (t1.matchDocs == t2.matchDocs) && 61 (t1.termFreq == t2.termFreq)); 59 62 } 60 63 … … 442 445 unsigned long termDocFreq = 0; 443 446 unsigned long lastLevelDocNum = 0; 444 447 unsigned long overallwordfreq = 0; 445 448 446 449 while (termDataI < termDataSize) { … … 463 466 if (needRanks) 464 467 termDocFreq += termData.fragFreqs[termDataI]; 468 overallwordfreq += termData.fragFreqs[termDataI]; 465 469 } 466 470 … … 484 488 termFreqData.stemMethod = stemMethod; 485 489 termFreqData.matchDocs = termData.matchDocs; 490 termFreqData.termFreq = overallwordfreq; 486 491 result.termFreqs.push_back (termFreqData); 487 492 } … … 517 522 unsigned long termDocFreq = 0; 518 523 unsigned long lastLevelDocNum = 0; 519 524 unsigned long overallwordfreq = 0; 520 525 unsigned long resultI = 0; 521 526 unsigned long resultSize = result.docs.size(); … … 552 557 if (needRanks) 553 558 termDocFreq += termData.fragFreqs[termDataI]; 559 overallwordfreq += termData.fragFreqs[termDataI]; 554 560 } 555 561 556 562 termDataI++; 557 } 563 } // while 558 564 559 565 if (lastLevelDocNum > 0) { … … 590 596 termFreqData.stemMethod = stemMethod; 591 597 termFreqData.matchDocs = termData.matchDocs; 598 termFreqData.termFreq = overallwordfreq; 592 599 result.termFreqs.push_back (termFreqData); 593 600 } -
trunk/gsdl/src/mgpp/text/Terms.h
r927 r1124 58 58 UCArray term; // unstemmed term 59 59 int stemMethod; 60 unsigned long matchDocs; // tf for level 61 60 unsigned long matchDocs; // tf for level - num levels 61 // containing this term 62 unsigned long termFreq; // overall term freq - num words that 63 // are this term 62 64 void Clear (); 63 65 TermFreqData () { Clear (); }
Note:
See TracChangeset
for help on using the changeset viewer.