Omnibox: Create Field Trial for HQP to Ignore Mid-Word Matches

chrome/browser/history/scored_history_match.h

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #ifndef CHROME_BROWSER_HISTORY_SCORED_HISTORY_MATCH_H_
 #define CHROME_BROWSER_HISTORY_SCORED_HISTORY_MATCH_H_
 
 #include <map>
 #include <set>
 #include <vector>
@@ skipping 20 matching lines @@
   // If the row does not qualify the raw score will be 0. |bookmark_service| is
   // used to determine if the match's URL is referenced by any bookmarks.
   ScoredHistoryMatch(const URLRow& row,
                      const string16& lower_string,
                      const String16Vector& terms_vector,
                      const RowWordStarts& word_starts,
                      const base::Time now,
                      BookmarkService* bookmark_service);
   ~ScoredHistoryMatch();
 
-  // Calculates a component score based on position, ordering and total
-  // substring match size using metrics recorded in |matches|. |max_length|
-  // is the length of the string against which the terms are being searched.
+  // Calculates a component score based on position, ordering, word
+  // boundaries, and total substring match size using metrics recorded
+  // in |matches| and |word_starts|. |max_length| is the length of
+  // the string against which the terms are being searched.
   static int ScoreComponentForMatches(const TermMatches& matches,

[Bart N., 2012/11/29 17:32:26]

matches -> provided_matches to make it consistent with your implementation.

[Mark P, 2012/11/29 19:36:48]

Done.

+                                      const WordStarts& word_starts,
                                       size_t max_length);
 
+  // Given a set of term matches |matches| and word boundaries |word_starts|,
+  // fills in |matches_at_word_boundaries| with only the matches in |matches|
+  // that are at word boundaries.
+  static void MakeTermMatchesOnlyAtWordBoundaries(
+      const TermMatches& matches,
+      const WordStarts& word_starts,
+      TermMatches* matches_at_word_boundaries);
+
   // Converts a raw value for some particular scoring factor into a score
   // component for that factor.  The conversion function is piecewise linear,
   // with input values provided in |value_ranks| and resulting output scores
   // from |kScoreRank| (mathematically, f(value_rank[i]) = kScoreRank[i]).  A
   // score cannot be higher than kScoreRank[0], and drops directly to 0 if
   // lower than kScoreRank[3].
   //
   // For example, take |value| == 70 and |value_ranks| == { 100, 50, 30, 10 }.
   // Because 70 falls between ranks 0 (100) and 1 (50), the score is given by
   // the linear function:
@@ skipping 38 matching lines @@
 
   // Returns a popularity score based on |typed_count| and
   // |visit_count|.
   static float GetPopularityScore(int typed_count,
                                   int visit_count);
 
   // Sets use_new_scoring based on command line flags and/or
   // field trial state.
   static void InitializeNewScoringField();
 
+  // Sets only_count_matches_at_word_boundaries based on the field trial state.
+  static void InitializeOnlyCountMatchesAtWordBoundariesField();
+
   // Sets also_do_hup_like_scoring based on the field trial state.
   static void InitializeAlsoDoHUPLikeScoringField();
 
   // End of functions used only in "new" scoring --------------------------
 
   // An interim score taking into consideration location and completeness
   // of the match.
   int raw_score;
   TermMatches url_matches;  // Term matches within the URL.
   TermMatches title_matches;  // Term matches within the page title.
@@ skipping 12 matching lines @@
   // scores.  |raw_term_score_to_topicality_score| is a simple array
   // mapping how raw terms scores (a weighted sum of the number of
   // hits for the term, weighted by how important the hit is:
   // hostname, path, etc.) to the topicality score we should assign
   // it.  This allows easy lookups of scores without requiring math.
   // This is initialized upon first use of GetTopicalityScore(),
   // which calls FillInTermScoreToTopicalityScoreArray().
   static const int kMaxRawTermScore = 30;
   static float* raw_term_score_to_topicality_score;
 
-  // Allows us to determing setting for use_new_scoring_ only once.
+  // Allows us to determing setting for use_new_scoring_ and other startup

[Bart N., 2012/11/29 17:32:26]

-> |use_new_scoring| (extra _)?

[Mark P, 2012/11/29 19:36:48]

Rephrased entirely.

+  // calculations only once.
   static bool initialized_;
 
   // Whether to use new-scoring or old-scoring.  Set in the
   // constructor by examining command line flags and field trial
   // state.  Note that new-scoring has to do with a new version of the
   // ordinary scoring done here.  It has nothing to do with and no
   // affect on HistoryURLProvider-like scoring that can happen in this
   // class as well (see boolean below).
   static bool use_new_scoring;
 
+  // If true, we ignore all matches that are in the middle of a word.
+  static bool only_count_matches_at_word_boundaries;
+
   // If true, assign raw scores to be max(whatever it normally would be,
   // a score that's similar to the score HistoryURL provider would assign).
   // This variable is set in the constructor by examining the field trial
   // state.
   static bool also_do_hup_like_scoring;
 };
 typedef std::vector<ScoredHistoryMatch> ScoredHistoryMatches;
 
 }  // namespace history
 
 #endif  // CHROME_BROWSER_HISTORY_SCORED_HISTORY_MATCH_H_