Microsoft Technology Licensing, LLC

Rong Xiao

Lei Zhang

Xiaoou Tang

Jiangming Yang

Xingrong Chen

Rui Cai

Fang Wen

Changhu Wang

Qiang Hao

Wei-Ying Ma

Hong-Jiang Zhang

Zhouchen Lin

Mingjing Li

Weiwei Zhang

Qiong Yang

Yan-Feng Sun

Deli Zhao

Long Zhu

Xiao-Ou Tang

A location extraction component analyzes a set of travelogues to identify all of the locations mentioned therein. A co-occurrence extraction component computes co-occurrence values for the identified locations. When the identity of a specified location is received, suggested locations for the specified location are identified through the use of the co-occurrence values. A map is displayed that encompasses an area including the specified location and the suggested locations. The map might include indicators for the specified location and for each of the suggested locations. Attributes of the indicators, such as their size or color, can be modified based upon the co-occurrence value associated with the corresponding suggested location.

Travelogue locating mining for travel suggestion

Described herein is a technology that facilitates efficient automated mining of topic-related aspects of user-generated content based on automated analysis of the user-generated content. Locations are automatically learned based on dividing documents into document segments, and decomposing the segments into local topics and global topics. Techniques are described that facilitate automatically extracting snippets. These techniques include, for example, computer annotating travelogues with learned tags and images, performing topic learning to obtain an interest model, performing location matching based on the interest model, calculating geographic and semantic relevance scores, ranking snippets based on the geographic and semantic relevance scores, and searching snippets with a 'location+context term' query.

Snippet extraction and ranking

Concepts and technologies are described herein for identifying location names within document text. Through an implementation of the concepts and technologies presented herein, functionality can be provided for identifying location names within articles, websites, travelogues, or other such documents. For instance, documents containing the names of cities, regions, countries, landmarks, or other locations may be associated with those locations. The location names may be unambiguously identified even when the location names may also have common word meanings that are not location associated or when the location name may be associated with more than one location.

Identifying location names within document text

Concepts and technologies are described herein for mining location contexts within document text. Through an implementation of the concepts and technologies presented herein, functionality can be provided for location context mining within articles, websites, travelogues, or other such documents. A location context is a concept associated with a specific location. For example, the contexts 'beach' and 'hula' are associated with Hawaii. Similarly, &#x201c;glacier&#x201d; and &#x201c;polar bear&#x201d; are contexts associated with Alaska. Location context mining can automatically discover locations and location contexts by mining information from a set of documents. User interfaces to support queries of the mined information are also presented herein.

Location context mining

Described herein is a technology that facilitates efficient automated mining of topic-related aspects of user generated content based on automated analysis of the user generated content. Locations are automatically learned based on dividing documents into document segments, and decomposing the segments into local topics and global topics. Techniques described herein include, for example, computer annotating travelogues with learned tags, performing topic learning to obtain an interest model, and performing location matching based on the interest model.

Mining topic-related aspects from user generated content

A map user interface control provides functionality for displaying a map in conjunction with the display of a Web page. The map control operates in combination with a location extraction component that analyzes the contents of the Web page to identify locations mentioned therein. Once the location extraction component has identified the locations mentioned in the Web page, a map is generated that encompasses the locations identified in the Web page. Once the map has been generated, the map control displays the map in conjunction with the display of the Web page. The map might include visual indicators corresponding to the locations mentioned in the Web page. The map might also include visual indicators corresponding to other locations near the locations identified in the Web page that have been identified using co-occurrence values generated through an analysis of a set of travelogues.

Travelogue-based contextual map generation

A location extraction component analyzes a set of travelogues to identify locations mentioned therein. A co-occurrence extraction component computes co-occurrence values for the identified locations. When a request to generate a travel route from a starting location to an ending location is received, suggested locations on or near the travel route are identified through the use of the co-occurrence values. A suggested travel route is then generated that passes through the starting location, the ending location, and the suggested locations. A map may be displayed showing the starting location, the ending location, the suggested locations, and the suggested travel route.

Travelogue-based travel route planning

An exemplary method for extracting discriminant feature of samples includes providing data for samples in a multidimensional space; based on the data, computing local similarities for the samples; mapping the local similarities to weights; based on the mapping, formulating an inter-class scatter matrix and an intra-class scatter matrix; and based on the matrices, maximizing the ratio of inter-class scatter to intra-class scatter for the samples to provide discriminate features of the samples. Such a method may be used for classifying samples, recognizing patterns, or other tasks. Various other methods, devices, system, etc., are also disclosed.

Linear laplacian discrimination for feature extraction

An interactive photo annotation method uses clustering based on facial similarities to improve annotation experience. The method uses a face recognition algorithm to extract facial features of a photo album and cluster the photos into multiple face groups based on facial similarity. The method annotates a face group collectively using annotations, such as name identifiers, in one operation. The method further allows merging and splitting of face groups. Special graphical user interfaces, such as displays in a group view area and a thumbnail area and drag-and-drop features, are used to further improve the annotation experience.

Interactive photo annotation based on face clustering

An object detector that includes a number of weak classifiers can be trained using a subset (a 'working set') of training data instead of all of the training data. The working set can be updated so that, for example, it remains representative of the training data. A decision to update the working set may be made based on the false positive sample rate&#x2014;if that rate falls below a threshold value, an update of the working set can be triggered.

Object detector trained using a working set of training data

Systems and methods are described for robust online face tracking. In one implementation, a system derives multiple resolutions of each video frame of a video sequence portraying movement of a visual object. The system tracks movement of the visual object in a low resolution as input for tracking the visual object in a higher resolution. The system can greatly reduce jitter while maintaining an ability to reliably track fast-moving visual objects.

Robust online face tracking

Systems and methods are described for a face annotation framework with partial clustering and interactive labeling. In one implementation, an exemplary system automatically groups some images of a collection of images into clusters, each cluster mainly including images that contain a person's face associated with that cluster. After an initial user-labeling of each cluster with the person's name or other label, in which the user may also delete/label images that do not belong in the cluster, the system iteratively proposes subsequent clusters for the user to label, proposing clusters of images that when labeled, produce a maximum information gain at each iteration and minimize the total number of user interactions for labeling the entire collection of images.

Face annotation framework with partial clustering and interactive labeling

An efficient, effective and at times superior object detection and/or recognition (ODR) function may be built from a set of Bayesian stumps. Bayesian stumps may be constructed for each feature and object class, and the ODR function may be constructed from the subset of Bayesian stumps that minimize Bayesian error for a particular object class. That is, Bayesian error may be utilized as a feature selection measure for the ODR function. Furthermore, Bayesian stumps may be efficiently implemented as lookup tables with entries corresponding to unequal intervals of feature histograms. Interval widths and entry values may be determined so as to minimize Bayesian error, yielding Bayesian stumps that are optimal in this respect.

Object detection and recognition with bayesian boosting

Face detection techniques are provided that use a multiple-stage face detection algorithm. An exemplary three-stage algorithm includes a first stage that applies linear-filtering to enhance detection performance by removing many non-face-like portions within an image, a second stage that uses a boosting chain that is adopted to combine boosting classifiers within a hierarchy 'chain' structure, and a third stage that performs post-filtering using image pre-processing, SVM-filtering and color-filtering to refine the final face detection prediction. In certain further implementations, the face detection techniques include a two-level hierarchy in-plane pose estimator to provide a rapid multi-view face detector that further improves the accuracy and robustness of face detection.

Robust multi-view face detection methods and apparatuses

Methods and systems are provided for selecting features that will be used to recognize faces. Three-dimensional models are used to synthesize a database of virtual face images. The virtual face images cover wide appearance variations, different poses, different lighting conditions and expression changes. A joint boosting algorithm is used to identify discriminative features by selecting features from the plurality of virtual images such that the identified discriminative features are independent of the other images included in the database.

Growing community of inventors

Beijing, China

Rong Xiao