OLD | NEW |
---|---|
1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "components/omnibox/browser/scored_history_match.h" | 5 #include "components/omnibox/browser/scored_history_match.h" |
6 | 6 |
7 #include <math.h> | 7 #include <math.h> |
8 | 8 |
9 #include <algorithm> | 9 #include <algorithm> |
10 #include <vector> | 10 #include <vector> |
(...skipping 91 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
102 } | 102 } |
103 | 103 |
104 } // namespace | 104 } // namespace |
105 | 105 |
106 // static | 106 // static |
107 bool ScoredHistoryMatch::also_do_hup_like_scoring_; | 107 bool ScoredHistoryMatch::also_do_hup_like_scoring_; |
108 float ScoredHistoryMatch::bookmark_value_; | 108 float ScoredHistoryMatch::bookmark_value_; |
109 float ScoredHistoryMatch::typed_value_; | 109 float ScoredHistoryMatch::typed_value_; |
110 bool ScoredHistoryMatch::fix_few_visits_bug_; | 110 bool ScoredHistoryMatch::fix_few_visits_bug_; |
111 bool ScoredHistoryMatch::frequency_uses_sum_; | 111 bool ScoredHistoryMatch::frequency_uses_sum_; |
112 OmniboxFieldTrial::NumMatchesMultipliers* | |
113 ScoredHistoryMatch::num_matches_to_frequency_multiplier_ = nullptr; | |
112 size_t ScoredHistoryMatch::max_visits_to_score_; | 114 size_t ScoredHistoryMatch::max_visits_to_score_; |
113 bool ScoredHistoryMatch::allow_tld_matches_; | 115 bool ScoredHistoryMatch::allow_tld_matches_; |
114 bool ScoredHistoryMatch::allow_scheme_matches_; | 116 bool ScoredHistoryMatch::allow_scheme_matches_; |
115 size_t ScoredHistoryMatch::num_title_words_to_allow_; | 117 size_t ScoredHistoryMatch::num_title_words_to_allow_; |
116 bool ScoredHistoryMatch::hqp_experimental_scoring_enabled_; | 118 bool ScoredHistoryMatch::hqp_experimental_scoring_enabled_; |
117 | 119 |
118 // Default topicality threshold. See GetTopicalityScore() for details. | 120 // Default topicality threshold. See GetTopicalityScore() for details. |
119 float ScoredHistoryMatch::topicality_threshold_ = 0.8f; | 121 float ScoredHistoryMatch::topicality_threshold_ = 0.8f; |
120 | 122 |
121 // Default HQP relevance buckets. See GetFinalRelevancyScore() for more details | 123 // Default HQP relevance buckets. See GetFinalRelevancyScore() for more details |
122 // on these numbers. | 124 // on these numbers. |
123 char ScoredHistoryMatch::hqp_relevance_buckets_str_[] = | 125 char ScoredHistoryMatch::hqp_relevance_buckets_str_[] = |
124 "0.0:400,1.5:600,5.0:900,10.5:1203,15.0:1300,20.0:1399"; | 126 "0.0:400,1.5:600,5.0:900,10.5:1203,15.0:1300,20.0:1399"; |
125 std::vector<ScoredHistoryMatch::ScoreMaxRelevance>* | 127 std::vector<ScoredHistoryMatch::ScoreMaxRelevance>* |
126 ScoredHistoryMatch::hqp_relevance_buckets_ = nullptr; | 128 ScoredHistoryMatch::hqp_relevance_buckets_ = nullptr; |
127 | 129 |
128 ScoredHistoryMatch::ScoredHistoryMatch() | 130 ScoredHistoryMatch::ScoredHistoryMatch() |
129 : ScoredHistoryMatch(history::URLRow(), | 131 : ScoredHistoryMatch(history::URLRow(), |
130 VisitInfoVector(), | 132 VisitInfoVector(), |
131 base::string16(), | 133 base::string16(), |
132 String16Vector(), | 134 String16Vector(), |
133 WordStarts(), | 135 WordStarts(), |
134 RowWordStarts(), | 136 RowWordStarts(), |
135 false, | 137 false, |
136 base::Time::Max()) { | 138 1, |
137 } | 139 base::Time::Max()) {} |
138 | 140 |
139 ScoredHistoryMatch::ScoredHistoryMatch( | 141 ScoredHistoryMatch::ScoredHistoryMatch( |
140 const history::URLRow& row, | 142 const history::URLRow& row, |
141 const VisitInfoVector& visits, | 143 const VisitInfoVector& visits, |
142 const base::string16& lower_string, | 144 const base::string16& lower_string, |
143 const String16Vector& terms_vector, | 145 const String16Vector& terms_vector, |
144 const WordStarts& terms_to_word_starts_offsets, | 146 const WordStarts& terms_to_word_starts_offsets, |
145 const RowWordStarts& word_starts, | 147 const RowWordStarts& word_starts, |
146 bool is_url_bookmarked, | 148 bool is_url_bookmarked, |
149 size_t num_matching_pages, | |
147 base::Time now) | 150 base::Time now) |
148 : HistoryMatch(row, 0, false, false), raw_score(0) { | 151 : HistoryMatch(row, 0, false, false), raw_score(0) { |
149 // NOTE: Call Init() before doing any validity checking to ensure that the | 152 // NOTE: Call Init() before doing any validity checking to ensure that the |
150 // class is always initialized after an instance has been constructed. In | 153 // class is always initialized after an instance has been constructed. In |
151 // particular, this ensures that the class is initialized after an instance | 154 // particular, this ensures that the class is initialized after an instance |
152 // has been constructed via the no-args constructor. | 155 // has been constructed via the no-args constructor. |
153 ScoredHistoryMatch::Init(); | 156 ScoredHistoryMatch::Init(); |
154 | 157 |
155 // Figure out where each search term appears in the URL and/or page title | 158 // Figure out where each search term appears in the URL and/or page title |
156 // so that we can score as well as provide autocomplete highlighting. | 159 // so that we can score as well as provide autocomplete highlighting. |
(...skipping 96 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
253 // the prefix that makes it inlineable may be empty. | 256 // the prefix that makes it inlineable may be empty. |
254 likely_can_inline = true; | 257 likely_can_inline = true; |
255 innermost_match = (best_inlineable_prefix->num_components == | 258 innermost_match = (best_inlineable_prefix->num_components == |
256 best_prefix->num_components); | 259 best_prefix->num_components); |
257 } | 260 } |
258 } | 261 } |
259 } | 262 } |
260 | 263 |
261 const float topicality_score = GetTopicalityScore( | 264 const float topicality_score = GetTopicalityScore( |
262 terms_vector.size(), url, terms_to_word_starts_offsets, word_starts); | 265 terms_vector.size(), url, terms_to_word_starts_offsets, word_starts); |
263 const float frequency_score = GetFrequency(now, is_url_bookmarked, visits); | 266 const float frequency_score = |
267 GetFrequency(now, is_url_bookmarked, visits, num_matching_pages); | |
264 raw_score = base::saturated_cast<int>(GetFinalRelevancyScore( | 268 raw_score = base::saturated_cast<int>(GetFinalRelevancyScore( |
265 topicality_score, frequency_score, *hqp_relevance_buckets_)); | 269 topicality_score, frequency_score, *hqp_relevance_buckets_)); |
266 | 270 |
267 if (also_do_hup_like_scoring_ && likely_can_inline) { | 271 if (also_do_hup_like_scoring_ && likely_can_inline) { |
268 // HistoryURL-provider-like scoring gives any match that is | 272 // HistoryURL-provider-like scoring gives any match that is |
269 // capable of being inlined a certain minimum score. Some of these | 273 // capable of being inlined a certain minimum score. Some of these |
270 // are given a higher score that lets them be shown in inline. | 274 // are given a higher score that lets them be shown in inline. |
271 // This test here derives from the test in | 275 // This test here derives from the test in |
272 // HistoryURLProvider::PromoteMatchForInlineAutocomplete(). | 276 // HistoryURLProvider::PromoteMatchForInlineAutocomplete(). |
273 const bool promote_to_inline = | 277 const bool promote_to_inline = |
(...skipping 131 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
405 if (initialized) | 409 if (initialized) |
406 return; | 410 return; |
407 | 411 |
408 initialized = true; | 412 initialized = true; |
409 also_do_hup_like_scoring_ = OmniboxFieldTrial::HQPAlsoDoHUPLikeScoring(); | 413 also_do_hup_like_scoring_ = OmniboxFieldTrial::HQPAlsoDoHUPLikeScoring(); |
410 bookmark_value_ = OmniboxFieldTrial::HQPBookmarkValue(); | 414 bookmark_value_ = OmniboxFieldTrial::HQPBookmarkValue(); |
411 typed_value_ = OmniboxFieldTrial::HQPTypedValue(); | 415 typed_value_ = OmniboxFieldTrial::HQPTypedValue(); |
412 max_visits_to_score_ = OmniboxFieldTrial::HQPMaxVisitsToScore(); | 416 max_visits_to_score_ = OmniboxFieldTrial::HQPMaxVisitsToScore(); |
413 frequency_uses_sum_ = OmniboxFieldTrial::HQPFreqencyUsesSum(); | 417 frequency_uses_sum_ = OmniboxFieldTrial::HQPFreqencyUsesSum(); |
414 fix_few_visits_bug_ = OmniboxFieldTrial::HQPFixFewVisitsBug(); | 418 fix_few_visits_bug_ = OmniboxFieldTrial::HQPFixFewVisitsBug(); |
419 num_matches_to_frequency_multiplier_ = | |
420 new OmniboxFieldTrial::NumMatchesMultipliers(); | |
Peter Kasting
2016/12/01 07:07:53
This leaks. If you want a leaky object of class t
Mark P
2016/12/04 01:06:42
Still TODO.
Mark P
2016/12/06 21:02:47
I don't understand how this macro should be used i
Peter Kasting
2016/12/06 21:41:27
There are two ways to do this I can think of:
(1)
Mark P
2016/12/08 00:21:31
Thanks for the explanation. I'll do something lik
| |
421 OmniboxFieldTrial::HQPGetNumMatchesMultipliers( | |
422 num_matches_to_frequency_multiplier_); | |
415 allow_tld_matches_ = OmniboxFieldTrial::HQPAllowMatchInTLDValue(); | 423 allow_tld_matches_ = OmniboxFieldTrial::HQPAllowMatchInTLDValue(); |
416 allow_scheme_matches_ = OmniboxFieldTrial::HQPAllowMatchInSchemeValue(); | 424 allow_scheme_matches_ = OmniboxFieldTrial::HQPAllowMatchInSchemeValue(); |
417 num_title_words_to_allow_ = OmniboxFieldTrial::HQPNumTitleWordsToAllow(); | 425 num_title_words_to_allow_ = OmniboxFieldTrial::HQPNumTitleWordsToAllow(); |
418 | 426 |
419 InitRawTermScoreToTopicalityScoreArray(); | 427 InitRawTermScoreToTopicalityScoreArray(); |
420 InitDaysAgoToRecencyScoreArray(); | 428 InitDaysAgoToRecencyScoreArray(); |
421 InitHQPExperimentalParams(); | 429 InitHQPExperimentalParams(); |
422 } | 430 } |
423 | 431 |
424 float ScoredHistoryMatch::GetTopicalityScore( | 432 float ScoredHistoryMatch::GetTopicalityScore( |
(...skipping 149 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
574 // Lookup the score in days_ago_to_recency_score, treating | 582 // Lookup the score in days_ago_to_recency_score, treating |
575 // everything older than what we've precomputed as the oldest thing | 583 // everything older than what we've precomputed as the oldest thing |
576 // we've precomputed. The std::max is to protect against corruption | 584 // we've precomputed. The std::max is to protect against corruption |
577 // in the database (in case last_visit_days_ago is negative). | 585 // in the database (in case last_visit_days_ago is negative). |
578 return days_ago_to_recency_score[std::max( | 586 return days_ago_to_recency_score[std::max( |
579 std::min(last_visit_days_ago, kDaysToPrecomputeRecencyScoresFor - 1), 0)]; | 587 std::min(last_visit_days_ago, kDaysToPrecomputeRecencyScoresFor - 1), 0)]; |
580 } | 588 } |
581 | 589 |
582 float ScoredHistoryMatch::GetFrequency(const base::Time& now, | 590 float ScoredHistoryMatch::GetFrequency(const base::Time& now, |
583 const bool bookmarked, | 591 const bool bookmarked, |
584 const VisitInfoVector& visits) const { | 592 const VisitInfoVector& visits, |
593 const size_t num_matching_pages) const { | |
585 // Compute the weighted sum of |value_of_transition| over the last at most | 594 // Compute the weighted sum of |value_of_transition| over the last at most |
586 // |max_visits_to_score_| visits, where each visit is weighted using | 595 // |max_visits_to_score_| visits, where each visit is weighted using |
587 // GetRecencyScore() based on how many days ago it happened. | 596 // GetRecencyScore() based on how many days ago it happened. |
588 float summed_visit_points = 0; | 597 float summed_visit_points = 0; |
589 auto visits_end = | 598 auto visits_end = |
590 visits.begin() + std::min(visits.size(), max_visits_to_score_); | 599 visits.begin() + std::min(visits.size(), max_visits_to_score_); |
591 // Visits should be in newest to oldest order. | 600 // Visits should be in newest to oldest order. |
592 DCHECK(std::adjacent_find( | 601 DCHECK(std::adjacent_find( |
593 visits.begin(), visits_end, | 602 visits.begin(), visits_end, |
594 [](const history::VisitInfo& a, const history::VisitInfo& b) { | 603 [](const history::VisitInfo& a, const history::VisitInfo& b) { |
595 return a.first < b.first; | 604 return a.first < b.first; |
596 }) == visits_end); | 605 }) == visits_end); |
597 for (auto i = visits.begin(); i != visits_end; ++i) { | 606 for (auto i = visits.begin(); i != visits_end; ++i) { |
598 const bool is_page_transition_typed = | 607 const bool is_page_transition_typed = |
599 ui::PageTransitionCoreTypeIs(i->second, ui::PAGE_TRANSITION_TYPED); | 608 ui::PageTransitionCoreTypeIs(i->second, ui::PAGE_TRANSITION_TYPED); |
600 float value_of_transition = is_page_transition_typed ? typed_value_ : 1; | 609 float value_of_transition = is_page_transition_typed ? typed_value_ : 1; |
601 if (bookmarked) | 610 if (bookmarked) |
602 value_of_transition = std::max(value_of_transition, bookmark_value_); | 611 value_of_transition = std::max(value_of_transition, bookmark_value_); |
603 const float bucket_weight = GetRecencyScore((now - i->first).InDays()); | 612 const float bucket_weight = GetRecencyScore((now - i->first).InDays()); |
604 summed_visit_points += (value_of_transition * bucket_weight); | 613 summed_visit_points += (value_of_transition * bucket_weight); |
605 } | 614 } |
606 if (frequency_uses_sum_) | 615 float frequency; |
607 return summed_visit_points; | 616 if (frequency_uses_sum_) { |
608 | 617 frequency = summed_visit_points; |
609 // Compute the average weighted value_of_transition and return it. | 618 } else { |
610 // Use |max_visits_to_score_| as the denominator for the average regardless of | 619 // Compute the average weighted value_of_transition and return it. |
611 // how many visits there were in order to penalize a match that has | 620 // Use |max_visits_to_score_| as the denominator for the average regardless |
612 // fewer visits than kMaxVisitsToScore. | 621 // of how many visits there were in order to penalize a match that has |
613 if (fix_few_visits_bug_) | 622 // fewer visits than kMaxVisitsToScore. |
614 return summed_visit_points / ScoredHistoryMatch::max_visits_to_score_; | 623 if (fix_few_visits_bug_) { |
615 return visits.size() * summed_visit_points / | 624 frequency = |
616 ScoredHistoryMatch::max_visits_to_score_; | 625 summed_visit_points / ScoredHistoryMatch::max_visits_to_score_; |
626 } else { | |
627 frequency = visits.size() * summed_visit_points / | |
628 ScoredHistoryMatch::max_visits_to_score_; | |
629 } | |
630 } | |
631 // Boost the score if applicable. | |
632 if (num_matches_to_frequency_multiplier_->find(num_matching_pages) != | |
633 num_matches_to_frequency_multiplier_->end()) | |
634 frequency *= (*num_matches_to_frequency_multiplier_)[num_matching_pages]; | |
635 return frequency; | |
Peter Kasting
2016/12/01 07:07:53
I don't love putting this here.
Part of this is c
Mark P
2016/12/04 01:06:42
I agree with your reasoning below. After some tho
| |
617 } | 636 } |
618 | 637 |
619 // static | 638 // static |
620 float ScoredHistoryMatch::GetFinalRelevancyScore( | 639 float ScoredHistoryMatch::GetFinalRelevancyScore( |
621 float topicality_score, | 640 float topicality_score, |
622 float frequency_score, | 641 float frequency_score, |
623 const std::vector<ScoreMaxRelevance>& hqp_relevance_buckets) { | 642 const std::vector<ScoreMaxRelevance>& hqp_relevance_buckets) { |
624 DCHECK(hqp_relevance_buckets.size() > 0); | 643 DCHECK(hqp_relevance_buckets.size() > 0); |
625 DCHECK_EQ(hqp_relevance_buckets[0].first, 0.0); | 644 DCHECK_EQ(hqp_relevance_buckets[0].first, 0.0); |
626 | 645 |
(...skipping 94 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
721 base::StringToDouble(it->first, &bucket.first); | 740 base::StringToDouble(it->first, &bucket.first); |
722 DCHECK(is_valid_intermediate_score); | 741 DCHECK(is_valid_intermediate_score); |
723 bool is_valid_hqp_score = base::StringToInt(it->second, &bucket.second); | 742 bool is_valid_hqp_score = base::StringToInt(it->second, &bucket.second); |
724 DCHECK(is_valid_hqp_score); | 743 DCHECK(is_valid_hqp_score); |
725 hqp_buckets->push_back(bucket); | 744 hqp_buckets->push_back(bucket); |
726 } | 745 } |
727 return true; | 746 return true; |
728 } | 747 } |
729 return false; | 748 return false; |
730 } | 749 } |
OLD | NEW |