Location Prediction Framework for Search Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search engines fail to accurately identify location-sensitive queries and rank relevant results due to the ambiguity of location information, both in queries and documents, as they often ignore implied locations and lack clear identification of relevant geographical locations.
Innovation Solution
A location prediction framework that applies location labels to documents and identifies location-sensitive queries by representing terms as term location vectors, using query-URL click data to construct a location-URL click graph, and training regression models to weight the importance of terms for location prediction, thereby improving the ranking of search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing search engines use traditional location recognition techniques (NER and Gazetteer), then explicit geographic terms can be identified, but implied locations and ambiguous location information are missed
Solution Approach 1:
The patent introduces term location vectors as an intermediary representation that bridges the gap between search terms and geographic locations. These vectors capture implicit location information by representing terms in a semantic space where location-related terms are positioned according to their geographic associations, enabling the system to infer locations even when explicit geographic terms are absent
Solution Approach 2:
The system transforms location identification from a discrete classification problem (using NER and Gazetteer) to a continuous vector space problem. By changing the representation parameter from binary location tags to continuous term location vectors, the system can capture nuanced implicit location information that traditional methods miss
2Ease of operation
If search engines rely on clear location tags in documents, then location-specific results can be identified, but documents with ambiguous or multiple locations are improperly ranked
Solution Approach 1:
The patent adds a new dimension to document representation by incorporating term location vectors alongside traditional content features. This dimensional extension allows the ranking system to simultaneously consider both content relevance and location relevance, enabling proper ranking of documents with ambiguous or multiple locations by evaluating their term vectors against query location sensitivity
3Measurement precision
If the system implements comprehensive location analysis for all queries, then location-sensitive queries are accurately identified, but computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary computation by pre-calculating term location vectors from training data and storing them for reuse. This preliminary action eliminates the need to recompute location relationships during query processing, reducing online computational complexity while maintaining high identification accuracy through the pre-built vector representations
Data Source
AI summary
A location prediction framework is described for applying location labels or tags to target documents and/or identifying location-sensitive queries. Terms in content and queries are represented by corresponding term locations vectors (TLVs) in which the term is represented as a weighted distribution across locations. Each element of a TLV represents a probability that the term corresponding to the TLV relates to a particular location. Predicted locations may be introduced as features to a ranking framework to improve the identification and ranking of search results for a given query.