# US 10,480,954 B2 · Vehicle routing guidance to an authoritative location for a point of interest > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,480,954 B2 - Document ID: US10480954B2 - Status: Granted - Kind: B2 - Application number: 15/606,555 - Filed: 2017-05-26 - Published: 2018-11-29 - Granted: 2019-11-19 - Earliest claimed: 2017-05-26 - Assignees: Uber Technologies, Inc. - Applicants: Uber Technologies, Inc. - Inventors: Chandan Prakash Sheth; Sheng Yang; Vasudev Parameswaran; Shivendra Pratap Singh; Jane Alam Jan - Web page: https://patentgrep.com/patent/US10480954B2 ## CPC - G01C 21/3476: {using point of interest [POI] information, e.g. a route passing visible POIs} - G01C 21/3679: {Retrieval, searching and output of POI information, e.g. hotels, restaurants, shops, filling stations, parking facilities (G01C21/3611 takes precedence)} - G06F 7/023: {adaptive, e.g. self learning} - G06F 7/14: Merging, i.e. combining at least two sets of record carriers each arranged in the same ordered sequence to produce a single set having the same ordered sequence - G06F 7/20: Comparing separate sets of record carriers arranged in the same sequence to determine whether at least some of the data in one set is identical with that in the other set or sets - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 20/00: Machine learning - G06N 20/20: Ensemble learning - G01C: MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY (measuring liquid level G01F; radio navigation, determining distance or velocity by use of propagation effects, e.g. Doppler effects, propagation time, of radio waves, analogous arrangements using other waves G01S) - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract An authoritative candidate is selected for determining a location of a point of interest (POI). Source data including name, address, and location for POIs is received from multiple data sources. The received data is normalized for ease of comparison, and coordinates for each candidate are compared to coordinates of other candidates to determine which candidate if any is an authoritative location for the POI. The candidate locations are compared using two models a metric-based scoring system and a machine learning model that may utilize a gradient boosted decision tree. The authoritative candidate can be used to render digital maps that include the POI. In addition, the authoritative candidate's location can be used to provide vehicle route guidance to the POI. ## Claims ### Claim 1 A computer-implemented method for providing vehicle routing guidance to a point of interest, comprising: receiving, by at least one processor from a plurality of data sources, point of interest data including at least one or more candidate locations for the point of interest and identifying information for the point of interest; obtaining map data including at least locations of road segments in an area surrounding the one or more candidate locations for the point of interest; for each of the one or more candidate locations: evaluating for the candidate location, based on the map data and the candidate location, a plurality of metrics, wherein one of the plurality of metrics is an across-the-road consensus metric; determining a metric score corresponding to each of the plurality of metrics; and aggregating the metric scores to calculate an aggregate score for the candidate location; selecting a first candidate location from the one or more candidate locations, the first candidate location having the highest aggregate score of the one or more candidate locations; providing vehicle routing guidance to the first candidate location. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, further comprising: for the one or more candidate locations: calculating a feature vector for the one or more candidate locations, the feature vector corresponding to the point of interest; applying a multiclass classifier to the calculated feature vector; classifying one of the one or more candidate locations as authoritative using the applied classifier; and calculating a confidence value associated with the classification of the candidate location; selecting the classified candidate location as a second candidate location from the one or more candidate locations; selecting an authoritative candidate location from the first selected candidate location and the second selected candidate location, based on the aggregate score of the first candidate location and the associated confidence value of the second candidate location; providing vehicle routing guidance to the selected authoritative candidate location. ### Claim 3 (depends on 2) The computer-implemented method of claim 2, wherein the multiclass classifier is a gradient boosted decision tree. ### Claim 4 (depends on 3) The computer-implemented method of claim 3, wherein the gradient boosted decision tree is trained by: retrieving training data containing a set of one or more candidate locations for each of a plurality of points of interest and a curated location for each of the plurality of points of interest, wherein each candidate location of each set of one or more candidate locations is associated with a different provider; obtaining map data including at least road segment data; for each of the plurality of points of interest in the training data: applying authority criteria to the set of one or more candidate locations associated with the point of interest; responsive to determining that a candidate location of the set of one or more candidate locations satisfies the authority criteria, based on the curated location, the set of one or more candidate locations associated with the point of interest, and the map data: selecting the candidate location as a target location for the point of interest; and responsive to determining that none of the set of one or more candidate locations satisfies the authority criteria, based on the curated location, the set of one or more candidate locations associated with the point of interest, and the map data: indicating that there is no target location for the point of interest; and applying a gradient boosting algorithm to train the gradient boosted decision tree that optimally classifies candidate locations in a set of one or more candidate locations as target locations or non-target locations for each point of interest in the training data, according to a loss function and a regularization function. ### Claim 5 (depends on 2) The computer-implemented method of claim 2, wherein the feature vector comprises a plurality of features including per-location features and per-pair features. ### Claim 6 (depends on 2) The computer-implemented method of claim 2, wherein the multiclass classifier includes multiple binary classifiers associated with each provider of candidate locations. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, wherein the plurality of metrics further includes at least one of: a building footprint consensus metric, a distance from the nearest road metric, a nearest-same-segment consensus metric, or a nearest segment popularity metric. ### Claim 8 A computer program product for providing vehicle routing guidance to a point of interest, the computer program product stored on a non-transitory computer-readable medium and including instructions that when executed cause a processor to carry out steps comprising: receiving, by at least one processor from a plurality of data sources, point of interest data including at least one or more candidate locations for the point of interest and identifying information for the point of interest; obtaining map data including at least locations of road segments in an area surrounding the one or more candidate locations for the point of interest; for the one or more candidate locations: calculating a feature vector for the one or more candidate locations, the feature vector corresponding to the point of interest; applying a multiclass classifier to the calculated feature vector; classifying one of the one or more candidate locations as authoritative using the applied classifier; and calculating a confidence value associated with the classification of the candidate location; selecting the classified candidate location as a first candidate location from the one or more candidate locations; providing vehicle routing guidance to the first candidate location. ### Claim 9 (depends on 8) The computer program product of claim 8, further comprising: for each of the one or more candidate locations: evaluating for the candidate location, based on the map data and the candidate location, a plurality of metrics; determining a metric score corresponding to each of the plurality of metrics; and aggregating the metric scores to calculate an aggregate score for the candidate location; selecting a second candidate location from the one or more candidate locations, the second candidate location having the highest aggregate score of the one or more candidate locations; selecting an authoritative candidate location from the first selected candidate location and the second selected candidate location, based on the aggregate score of the second candidate location and the associated confidence value of the first candidate location; providing vehicle routing guidance to the selected authoritative candidate location. ### Claim 10 (depends on 9) The computer program product of claim 9, wherein the plurality of metrics includes at least one of: an across-the-road consensus metric, a building footprint consensus metric, a distance from the nearest road metric, a nearest-same-segment consensus metric, or a nearest segment popularity metric. ### Claim 11 (depends on 8) The computer program product of claim 8, wherein the multiclass classifier is a gradient boosted decision tree. ### Claim 12 (depends on 11) The computer program product of claim 11, wherein the gradient boosted decision tree is trained by: retrieving training data containing a set of one or more candidate locations for each of a plurality of points of interest and a curated location for each of the plurality of points of interest, wherein each candidate location of each set of candidate locations is associated with a different provider; obtaining map data including at least road segment data; for each of the plurality of points of interest in the training data: applying authority criteria to the set of one or more candidate locations associated with the point of interest; responsive to determining that a candidate location of the set of one or more candidate locations satisfies the authority criteria, based on the curated location, the set of one or more candidate locations associated with the point of interest, and the map data: selecting the candidate location as a target location for the point of interest; and responsive to determining that none of the set of one or more candidate locations satisfies the authority criteria, based on the curated location, the set of one or more candidate locations associated with the point of interest, and the map data: indicating that there is no target location for the point of interest; and applying a gradient boosting algorithm to train the gradient boosted decision tree that optimally classifies candidate locations as target locations or non-target locations for each point of interest in the training data, according to a loss function and a regularization function. ### Claim 13 (depends on 8) The computer program product of claim 8, wherein the feature vector comprises a plurality of features including per-location features and per-pair features. ### Claim 14 (depends on 8) The computer program product of claim 8, wherein the multiclass classifier includes multiple binary classifiers associated with each provider of candidate locations. ### Claim 15 A system for providing vehicle routing guidance to a point of interest, comprising: at least one processor; a data normalizer module, executed by the processor, configured to receive from a plurality of data sources at least three candidate locations for the point of interest, at least one of the received candidate locations located across a road segment from another candidate; a point-of-interest location determination engine, executed by the processor, configured to perform the steps of: receiving, by at least one processor from the plurality of data sources, point of interest data including at least one or more candidate locations for the point of interest and identifying information for the point of interest; obtaining map data including at least locations of road segments in an area surrounding the one or more candidate locations for the point of interest; for the one or more candidate locations: calculating a feature vector for the one or more candidate locations, the feature vector corresponding to the point of interest; applying a multiclass classifier to the calculated feature vector; classifying one of the one or more candidate locations as authoritative using the applied classifier; and calculating a confidence value associated with the classification of the candidate location; selecting the classified candidate location as a first candidate location from the one or more candidate locations; providing vehicle routing guidance to the first candidate location. ### Claim 16 (depends on 15) The system of claim 15, further comprising: for each of the one or more candidate locations: evaluating for the candidate location, based on the map data and the candidate location, a plurality of metrics; determining a metric score corresponding to each of the plurality of metrics; and aggregating the metric scores to calculate an aggregate score for the candidate location; selecting a second candidate location from the one or more candidate locations, the second candidate location having the highest aggregate score of the one or more candidate locations; selecting an authoritative candidate location from the first selected candidate location and the second selected candidate location, based on the aggregate score of the second candidate location and the associated confidence value of the first candidate location; providing vehicle routing guidance to the selected authoritative candidate location. ### Claim 17 (depends on 16) The system of claim 16, wherein the plurality of metrics includes at least one of: an across-the-road consensus metric, a building footprint consensus metric, a distance from the nearest road metric, a nearest-same-segment consensus metric, or a nearest segment popularity metric. ### Claim 18 (depends on 15) The system of claim 15, wherein the multiclass classifier is a gradient boosted decision tree. ### Claim 19 (depends on 18) The system of claim 18, wherein the gradient boosted decision tree is trained by: retrieving training data containing a set of one or more candidate locations for each of a plurality of points of interest and a curated location for each of the plurality of points of interest, wherein each candidate location of each set of one or more candidate locations is associated with a different provider; obtaining map data including at least road segment data; for each of the plurality of points of interest in the training data: applying authority criteria to the set of one or more candidate locations associated with the point of interest; responsive to determining that a candidate location of the set of one or more candidate locations satisfies the authority criteria, based on the curated location, the set of one or more candidate locations associated with the point of interest, and the map data: selecting the candidate location as a target location for the point of interest; and responsive to determining that none of the set of one or more candidate locations satisfies the authority criteria, based on the curated location, the set of one or more candidate locations associated with the point of interest, and the map data: indicating that there is no target location for the point of interest; and applying a gradient boosting algorithm to train the gradient boosted decision tree that optimally classifies candidate locations as target locations or non-target locations for each point of interest in the training data, according to a loss function and a regularization function. ### Claim 20 (depends on 15) The system of claim 15, wherein the feature vector comprises a plurality of features including per-location features and per-pair features.