| Index: components/dom_distiller/core/page_features.h
|
| diff --git a/components/dom_distiller/core/page_features.h b/components/dom_distiller/core/page_features.h
|
| index 236796b542aecb860e3cb03311872183e50aec4f..a54d1cdfebf598a2d2858746e7e7848fe27b53b1 100644
|
| --- a/components/dom_distiller/core/page_features.h
|
| +++ b/components/dom_distiller/core/page_features.h
|
| @@ -10,19 +10,20 @@
|
| #include "base/values.h"
|
| #include "url/gurl.h"
|
|
|
| +class GURL;
|
| +
|
| namespace dom_distiller {
|
|
|
| // The length of the derived features vector.
|
| extern int kDerivedFeaturesCount;
|
|
|
| // The distillable page detector is a model trained on a list of numeric
|
| -// features derived from core more complex features of a webpage (like the
|
| -// body's .textContent). This derives the numeric features for a set of core
|
| -// features.
|
| +// features derived from features of a webpage (like body's number of elements
|
| +// ). This derives the numeric features form a set of core features.
|
| //
|
| // Note: It is crucial that these features are derived in the same way and are
|
| // in the same order as in the training pipeline. See //heuristics/distillable
|
| -// in the external DomDistillerJs repo.
|
| +// in the external DomDistiller repo.
|
| std::vector<double> CalculateDerivedFeatures(bool isOGArticle,
|
| const GURL& url,
|
| double numElements,
|
| @@ -37,6 +38,16 @@ std::vector<double> CalculateDerivedFeatures(bool isOGArticle,
|
| std::vector<double> CalculateDerivedFeaturesFromJSON(
|
| const base::Value* stringified_json);
|
|
|
| +std::vector<double> CalculateDerivedFeatures(
|
| + bool openGraph,
|
| + const GURL& url,
|
| + unsigned elementCount,
|
| + unsigned anchorCount,
|
| + unsigned formCount,
|
| + double mozScore,
|
| + double mozScoreAllSqrt,
|
| + double mozScoreAllLinear);
|
| +
|
| } // namespace dom_distiller
|
|
|
| #endif // COMPONENTS_DOM_DISTILLER_CORE_PAGE_FEATURES_H_
|
|
|