Method and apparatus for geographic document retrieval

摘要

A geographic document retrieval method (GDR) can be executed by a computer system to index, retrieve and rank geographical documents. Textual and spatial attributes of geographical documents are indexed separately using inverted index and spatial index, respectively. Spatial attributes of a document are represented as one or more contiguously closed regions of arbitrary shapes. Upon receiving an input query carrying a geographic representation of a location using arbitrary regions, the GDR method retrieves one or more documents by executing an overlap test between arbitrary regions from the query and the arbitrary regions associated with the documents.

主权项

1. A method of retrieving geographical documents performed by one or more computer systems coupled to a packet-based network, comprising:
generating a probabilistic representation of a location of a mobile device based on a request received from the packet-based network and historical data stored in a data store, the request being associated with the mobile device and including a location indicator and one or more other parameters, the probabilistic representation including multiple contiguously closed geographical regions of arbitrary shapes and probabilities of the mobile device being in respective ones of the multiple contiguously closed geographical regions; receiving a query having spatial clauses and at least one other clause, the spatial clauses representing the multiple contiguously closed geographical regions of arbitrary shapes, and the at least one other clause representing at least one of the one or more other parameters; retrieving first documents from geographical documents stored in a computer readable medium, the geographical documents each having associated therewith a spatial index and an inverted index, each of the first documents having its associated spatial index matching at least one of the spatial clauses; selecting second documents from the first documents, the second documents each having its associated inverted index matching the at least one other clause in the query; and selecting a third document from the second documents based at least on the probabilities of the mobile device being in the multiple contiguously closed geographical regions.