Abstract: A method and system obtaining positioning data from an object traveling on a plurality of routes; mapping the data into a plurality of points on a digital map; identifying points that are matched based on a distance and having a traveling direction consistent with the route directions; obtaining candidate transition points from the plurality of points on the digital map; aggregating the candidate transition points by applying a clustering algorithm to obtain a first cluster of points and a plurality of second clusters of points, verifying a confidence that the first cluster of points are transition points indicating a transition between the routes, and in response to the confidence being below a threshold confidence, classifying the first cluster of points as a first plurality of traveling points having a first direction and automatically adjusting the digital map at least in part based on the first cluster of points.

Abstract: A method and an apparatus for determining a location of a mobile device. The location of a mobile device is determined accurately according to information which includes call data records of the mobile device. By employing a partial ellipse integral model, two physical world factors are taken into consideration in reducing the location uncertainty in call data records. The factors include: spatiotemporal constraints of the device's movement in the physical world and the telecommunication cell area's geometry information, which increase the accuracy of determining the location of a mobile device.

Abstract: A method and system for obtaining a trajectory pattern of a route. One embodiment of the present invention provides a method for obtaining a trajectory pattern of a route. The method includes obtaining a set of points associated with the route, the set of points being obtained by at least one entity travelling along the route. A plurality of coverage areas covering a part of points in the set of points is determined. The coverage areas are connected based on main directions of the coverage areas to obtain the trajectory pattern related to the route. The main directions represent forward directions of the route in respective coverage areas. A corresponding system for obtaining a trajectory pattern of a route is described as well.

Abstract: A method of generating a predictor to classify data includes: training each of a plurality of first classifiers arranged in a first level on current training data; operating each classifier of the first level on the training data to generate a plurality of predictions; combining the current training data with the predictions to generated new training data; and training each of a plurality of second classifiers arranged in a second level on the new training data. The first classifiers are classifiers of different classifier types, respectively and the second classifiers are classifiers of the different classifier types, respectively.

Abstract: Provided is a mechanism for producing an uncertainty-based traffic congestion index, wherein the mechanism may comprise: obtaining a plurality of GPS data points; dividing the plurality of GPS data points into a plurality of variable sliding windows, wherein the dividing maximizes an amount of shape information in each of the plurality of variable sliding windows, performing a map matching process on the plurality of GPS data points as the GPS data points had been divided by the dividing; calculating a confidence value indicative of the map matching process; and producing the traffic congestion index, wherein the traffic congestion index is produced by taking into account the calculated confidence value indicative of the map matching process. In various embodiments, such a mechanism may be implemented via systems, methods and/or computer program products.

Abstract: A vehicle domain multi-level parallel buffering and context-based streaming data pre-processing system includes a first data processing level and a second data processing level. The first data processing level includes a first-level buffer configured to buffer data provided from a plurality of raw data streams output from a plurality of vehicles. The second data processing level includes an electronic task-queue-dictionary (TQD) module and a plurality of second-level data processing buffers. The TQD module is configured to create a plurality of tasks in response to receiving a serial data stream output from the first-level buffer. The TQD module is further configured to assign each task to a corresponding second-level buffer, and separate the serial data stream into individual data values that are delivered to a specific second-level buffer based on the task so as to generate a multi-level parallel context-based buffering operation.

Abstract: Location based on call detail record. The method includes: acquiring handover call detail record (HCDR) data and handover (HO) data corresponding to at least one cell transfer during a call; The HCDR data can include a start time, an end time, passed cells during the call; The HO data can include the HO time for each pair of adjacent cells among the passed cells during the call. The method can further include estimating the location of the subscriber based on the HCDR data and the HO data.

Abstract: Provided is a mechanism for producing an uncertainty-based traffic congestion index, wherein the mechanism may comprise: obtaining a plurality of GPS data points; dividing the plurality of GPS data points into a plurality of variable sliding windows, wherein the dividing maximizes an amount of shape information in each of the plurality of variable sliding windows, performing a map matching process on the plurality of GPS data points as the GPS data points had been divided by the dividing; calculating a confidence value indicative of the map matching process; and producing the traffic congestion index, wherein the traffic congestion index is produced by taking into account the calculated confidence value indicative of the map matching process. In various embodiments, such a mechanism may be implemented via systems, methods and/or computer program products.

Abstract: There is disclosed a method, comprising: determining a road segment group set that corresponds to a current time period, and a correspondence relationship between each road segment group in the road segment group set and a computing node of the cluster server, wherein the road segment group is a group of road segments which are grouped according to the number of computing nodes of the cluster server and the dissimilarity between divided road segments in a road network; in response to receiving multiple GPS sampled data from a vehicle, generating a map matching request according to the multiple GPS sampled data; and in response to generating a map matching request and obtaining a road segment to which the latest location of the vehicle belongs, sending the map matching request to a computing node that corresponds to a road segment group to which the obtained road segment belongs.

Abstract: A vehicle domain multi-level parallel buffering and context-based streaming data pre-processing system includes a first data processing level and a second data processing level. The first data processing level includes a first-level buffer configured to buffer data provided from a plurality of raw data streams output from a plurality of vehicles. The second data processing level includes an electronic task-queue-dictionary (TQD) module and a plurality of second-level data processing buffers. The TQD module is configured to create a plurality of tasks in response to receiving a serial data stream output from the first-level buffer. The TQD module is further configured to assign each task to a corresponding second-level buffer, and separate the serial data stream into individual data values that are delivered to a specific second-level buffer based on the task so as to generate a multi-level parallel context-based buffering operation.

Abstract: Location based on call detail record. The method includes: acquiring handover call detail record (HCDR) data and handover (HO) data corresponding to at least one cell transfer during a call; The HCDR data can include a start time, an end time, passed cells during the call; The HO data can include the HO time for each pair of adjacent cells among the passed cells during the call. The method can further include estimating the location of the subscriber based on the HCDR data and the HO data.

Abstract: In one embodiment, a computer-implemented method includes receiving training data including a plurality of records, each record having a plurality of attributes. The training data is horizontally parallelized across two or more processing elements. This horizontal parallelizing includes dividing the training data into two or more subsets of records; assigning each subset of records to a corresponding processing element of the two or more processing elements; transmitting each subset of records to its assigned processing element; and sorting, at the two or more processing elements, the two or more subsets of records to two or more candidate leaves of a decision tree. The output from horizontally parallelizing is converted into input for vertically parallelizing the training data. The training data is vertically parallelized across the two or more processing elements. The decision tree is grown based at least in part on the horizontally parallelizing, the converting, and the vertically parallelizing.

Abstract: In one embodiment, a computer-implemented method includes receiving training data including a plurality of records, each record having a plurality of attributes. The training data is horizontally parallelized across two or more processing elements. This horizontal parallelizing includes dividing the training data into two or more subsets of records; assigning each subset of records to a corresponding processing element of the two or more processing elements; transmitting each subset of records to its assigned processing element; and sorting, at the two or more processing elements, the two or more subsets of records to two or more candidate leaves of a decision tree. The output from horizontally parallelizing is converted into input for vertically parallelizing the training data. The training data is vertically parallelized across the two or more processing elements. The decision tree is grown based at least in part on the horizontally parallelizing, the converting, and the vertically parallelizing.

Abstract: Disclosed are a computer-implemented method for generating a data cube from data, a system and a computer program product. The method comprises selecting a candidate granularity from a plurality of candidate granularities determined for a dimension of the data cube, where a data distribution obtained in the selected candidate granularity satisfies a predetermined condition; and generating the data cube based on the selected candidate granularity for the dimension.

Abstract: A method and an apparatus for determining a location of a mobile device. The location of a mobile device is determined accurately according to information which includes call data records of the mobile device. By employing a partial ellipse integral model, two physical world factors are taken into consideration in reducing the location uncertainty in call data records. The factors include: spatiotemporal constraints of the device's movement in the physical world and the telecommunication cell area's geometry information, which increase the accuracy of determining the location of a mobile device.

Abstract: A vehicle domain multi-level parallel buffering and context-based streaming data pre-processing system includes a first data processing level and a second data processing level. The first data processing level includes a first-level buffer configured to buffer data provided from a plurality of raw data streams output from a plurality of vehicles. The second data processing level includes an electronic task-queue-dictionary (TQD) module and a plurality of second-level data processing buffers. The TQD module is configured to create a plurality of tasks in response to receiving a serial data stream output from the first-level buffer. The TQD module is further configured to assign each task to a corresponding second-level buffer, and separate the serial data stream into individual data values that are delivered to a specific second-level buffer based on the task so as to generate a multi-level parallel context-based buffering operation.

Abstract: A method of tagging a geographical area includes obtaining, with a processing device, attribute information and mobile tracking data of a plurality of mobile objects, wherein the mobile tracking data comprises sampling time and corresponding sampling point locations of the mobile objects; converting the mobile tracking data of the plurality of mobile objects into new mobile tracking data according to the correspondence relationship between the sampling time and a time slices, wherein the new mobile tracking data include time slices and corresponding sampling point locations; and obtaining a set of attribute information of at least one geographical area with respect to the time slices based on the new mobile tracking data, wherein the at least one geographical area is obtained by clustering the sampling point locations.

Abstract: A mechanism is provided in a data processing system for distributed tree learning. A source processing instance distributes data record instances to a plurality of model update processing items. The plurality of model update processing items determine candidate leaf splitting actions in a decision tree in parallel based on the data record instances. The plurality of model update processing items send the candidate leaf splitting actions to a plurality of conflict resolve processing items. The plurality of conflict resolve processing items identifies conflict leaf splitting actions. The plurality of conflict resolve processing items applies tree structure changes to the decision tree in the plurality of model update processing items.

Abstract: In one embodiment, a computer-implemented method includes receiving a global positioning system (GPS) location of a mobile device. Two or more road locations are determined as possible locations corresponding to the GPS location in a physical network of a set of roads. A first portion of a virtual network is built, by a computer processor, by expanding the virtual network from a node representing a prior road location to the two or more candidate road locations. A first candidate road location is selected as a current road location from among the two or more candidate road locations. A route of the mobile device is identified as including the prior road location and the first candidate road location. At least one of the two or more candidate road locations not selected as the current road location is excluded from future building of the virtual network. The virtual network is built dynamically.

Abstract: A mechanism is provided in a data processing system for distributed tree learning. A source processing instance distributes data record instances to a plurality of model update processing items. The plurality of model update processing items determine candidate leaf splitting actions in a decision tree in parallel based on the data record instances. The plurality of model update processing items send the candidate leaf splitting actions to a plurality of conflict resolve processing items. The plurality of conflict resolve processing items identifies conflict leaf splitting actions. The plurality of conflict resolve processing items applies tree structure changes to the decision tree in the plurality of model update processing items.