Identifies in-depth long-form articles for a query, scores them on combined topicality and in-depth quality, and surfaces them in a dedicated SERP module distinct from news, quick-answer, and standard web results.
Patent Overview
- Inventor
- Anand Shukla
- Assignee
- Google LLC
- Filed
- 2014-06-25
- Granted
- 2018-06-12
- Application Number
- US 14/315,008
The Challenge
Long-Form Content Loses To News And Quick Answers In Default Ranking
Standard ranking favors freshness and direct relevance, which crowds out comprehensive long-form articles even when the query intent benefits from them. A user searching for a broad topic often wants a deep explanatory article more than a 2-day-old news item. The system needs to identify in-depth articles separately, score them on combined topicality and quality, and surface them in a dedicated module so they don't have to compete head-to-head with news in the main ranking.
- Freshness Beats Depth In Standard Ranking — News articles with strong recency signals outrank comprehensive long-form pieces in standard ranking. Users with broad informational intent are under-served as a result.
- In-Depth Articles Have A Distinct Quality Signature — Long-form pieces share characteristics — word count, source authority, structural depth, references — that distinguish them from typical web pages. The system can detect this signature and treat in-depth content as its own class.
- Dedicated Module Avoids Direct Competition — Rather than ranking in-depth pieces against news in one list, the system carves out a separate module on the SERP. Long-form content competes within its module; news competes within its own.
- Per-Article Combined Scoring — Each candidate in-depth article needs two scores: topicality (relevance to the query) and in-depth-article-ness (quality as a long-form piece). Both combine into the document score that drives module selection.
Innovation
Topicality Plus In-Depth Score, Surface In Dedicated Module
The system first determines that in-depth articles should be provided for a given query (not every query benefits). For each candidate in-depth article, it obtains a topicality score and an in-depth-article score, then combines them into a document score. Top-scoring articles are selected and surfaced in a dedicated In-Depth Articles module on the search results page.
- Decide Whether In-Depth Module Applies — Classify the query for in-depth eligibility. Broad informational queries qualify; narrow transactional or news-pressed queries typically do not.
- Identify Candidate In-Depth Articles — From the indexed corpus, identify articles flagged as in-depth based on length, source authority, structural depth, and reference density.
- Score Topicality Per Article — Compute each candidate's topicality score relative to the current query. Standard relevance scoring applies.
- Score In-Depth Quality Per Article — Compute the in-depth-article score: how strongly the article exhibits the long-form quality signature.
- Combine Into Document Score — Combine topicality and in-depth scores into a single document score. The combination is the article's eligibility for module placement.
- Select Top Articles — Sort candidates by document score. Select the top N for the module.
- Surface In Dedicated Module — Render the selected articles in the In-Depth Articles module on the SERP, distinct from news and standard web results.
Two-Score Selection Plus Dedicated Module
The patent introduces both the dual-score evaluation and the separate-module presentation. Long-form content gets its own playing field rather than competing on the news-favored main ranking.
Different Content Classes, Different Surfaces
Long-form, news, quick-answer, and standard web results each have their own scoring and their own SERP modules. The architecture stops conflating fundamentally different content types into one list.
- Topicality Score — How relevant the article is to the query. Standard relevance computation.
- In-Depth-Article Score — How strongly the article exhibits long-form quality (length, depth, authority).
- Dedicated SERP Module — Articles surface in their own carousel/section, not the main ranking. Avoids head-to-head competition with news.
Technical Foundation
Scoring Components
Two independent scores combine into the document score that drives module selection.
- Topicality Score — Standard query-document relevance score. Computed using TF-IDF, BM25, semantic similarity, or learned ranking.
- In-Depth-Article Score — Per-article quality score for the long-form signature: length, references, source authority, structural depth, topical comprehensiveness.
- Document Score — Combined ranking score used to select articles for the module. Weighted combination of topicality and in-depth scores; weights tunable per query class.
Key Insight: Treating in-depth articles as a separate content class with its own scoring and its own SERP module is a structural decision that protects long-form content from being crowded out by freshness-favored news ranking. The dual-score model makes the in-depth class precisely defined: articles must be both topically relevant AND structurally long-form to qualify.
<\/section>What This Means for SEO
What This Means for SEO
In-depth article surfacing is one of the SERP features long-form publishers compete for. Knowing the dual-score mechanism informs how to position long content for module placement.
- Long-Form Content Needs Both Scores — An in-depth article must be topically relevant AND exhibit the in-depth quality signature. Long content on the wrong topic fails the topicality score; short or shallow content on the right topic fails the in-depth score. Both axes matter.
- Structural Depth Signals In-Depth Status — Articles with proper sectioning (H2/H3 hierarchy), references, deep paragraph structure, and topical comprehensiveness register as in-depth. Flat unstructured text of equal length does not.
- Length Plus Authority Plus References — The in-depth-article score combines length, source authority, and reference density. All three contribute. Long articles from unauthoritative sources or without references don't earn the score.
- Broad Informational Queries Trigger The Module — Not every query gets an in-depth module. Broad informational queries (concept explanations, topic surveys) trigger it; narrow transactional or news-led queries usually do not. Target queries where the module appears.
- Module Competition Is Separate From Main Ranking — Your in-depth article competes in the module against other in-depth articles, not against news and quick answers. The competitive set is smaller and quality-defined.
- Topical Comprehensiveness Beats Topical Specialization — The in-depth score rewards comprehensive coverage of a topic, not narrow specialization. Pillar pages that survey a topic broadly tend to score higher than narrow deep-dives on a sub-aspect.