Maps queries to a topic map and identifies trending subtopics within the topic, surfacing web documents from the user-selected trending subtopic to drive topic-aware exploration.
Patent Overview
- Inventor
- Anand Shukla
- Assignee
- Google LLC
- Filed
- 2018-02-26
- Granted
- 2019-08-29 (published application)
- Application Number
- US 15/905,797
The Challenge
Queries Are Single Points, Topics Are Spaces
A query represents the user's current entry point into a topic. But topics have internal structure: parent and child sub-topics, trending sub-themes, related angles. Returning documents only for the literal query misses the surrounding topical space the user might want to explore. The system needs to map queries onto a topic graph and offer trending sub-topics as exploration choices.
- Single-Query Retrieval Misses The Surrounding Topic Space — A user querying a broad topic often benefits from exposure to trending sub-topics they haven't named. Pure literal-query retrieval doesn't surface those branches.
- Trending Sub-Topics Reveal Live Interest — Within a parent topic, some sub-topics are trending (currently active, growing engagement) while others are stable or fading. Surfacing trending sub-topics matches user interest with current discourse.
- Need A Topic Map Substrate — The system needs to maintain a topic map: nodes representing topics, edges representing sub-topic relationships, with live trending signal attached to each node.
- User Selection Drives Retrieval Pivot — When the user picks a trending sub-topic, the retrieval pivots to documents on that sub-topic. The selection is the user's exploration intent.
Innovation
Map Query To Topic, Show Trending Sub-Topics, Pivot On Selection
The system identifies one or more trending sub-topics associated with the topic included in the user's query. The trending sub-topics are presented to the user. A selection is received. The system provides web documents associated with the selected trending sub-topic, pivoting retrieval to the user's chosen branch.
- Map Query To Topic — Identify the topic that the user's query falls under. Use the topic map's existing structure for the mapping.
- Identify Trending Sub-Topics — Within the parent topic, identify sub-topics currently trending. Trending detection uses recency, growth, and engagement signals attached to topic-map nodes.
- Present Sub-Topics To User — Surface the trending sub-topics as exploration choices alongside the main result set.
- Receive User Selection — User picks one of the trending sub-topics, signaling their exploration intent.
- Retrieve Documents For Selected Sub-Topic — Pivot retrieval to documents specifically associated with the selected trending sub-topic.
- Surface Sub-Topic Results — Display documents from the selected sub-topic. The user can refine further or pivot to a different branch.
Topic Map Plus Trending Signal
The topic map is the substrate; the trending signal is what makes selection useful. Users get topic-aware exploration without having to know the right sub-topic vocabulary in advance.
Topics Have Internal Structure
A topic is not a leaf — it contains sub-topics. The map captures this structure so retrieval can pivot within it.
- Topic Map — Nodes (topics) plus edges (sub-topic relationships). Maintained as a structured knowledge resource.
- Trending Signal Per Node — Each topic node carries a live trending measure. Used to filter and rank sub-topics for surfacing.
What This Means for SEO
What This Means for SEO
Topic-map navigation surfaces shape how users explore beyond their initial query. Understanding the map informs how to position content within topical structures.
- Sub-Topic Coverage Is Discovery Surface — If your content covers a trending sub-topic of a popular parent topic, it can be surfaced when users pivot to that sub-topic from their initial broader query.
- Trending Sub-Topics Are Time-Sensitive — What's trending changes. Content that catches a rising sub-topic early earns disproportionate exposure during the trend window.
- Pillar Plus Sub-Topic Pages Map Cleanly — A pillar page covering the parent topic plus child pages covering sub-topics aligns with the topic-map structure. The architecture mirrors how the system reads topical space.