104 results about "Clustering search" patented technology

Various aspects of the subject technology relate to systems, methods, and machine-readable media for clustering search results based on image composition. A system may, for each image in a set of images, determine a compositional vector representing one or more objects and corresponding locations within the image using a trained computer-operated convolutional neural network. The system may provide each image through a clustering algorithm to produce one or more clusters based on compositional similarity. The system may provide images from the set of images clustered by composition, in which the images include a different listing of images for each of the one or more clusters. The system may provide a prioritized listing of images responsive to a user search query, in which the prioritized listing of images includes a different listing of images for each cluster of compositional similarity based on the metadata of each image associated with the cluster.

The invention discloses a keyword extraction method based on a Seq2seq framework. The method comprises: creating a sequence model by utilizing a Seq2seq framework; introducing an attention mechanism,extracting features of keywords in the text; fusing a pointer network model and a Cover mechanism at a decoding end to improve the attention distribution of potential keywords; then using a softmax loss function to train a network model, and finally in a model prediction stage, using a Beam Search cluster search algorithm to generate a key word sequence with the maximum probability to serve as a key word result set to obtain appropriate key words. According to the method, deep semantics in the long text are well considered, the word distribution probability is calculated in combination with the context information context, the problem of repetition of low-frequency words and generative tasks is better solved, and the keyword extraction accuracy is improved.

An information resource taxonomy system, including a data collector for collecting information resources from a communications network; and a taxonomy generator for generating a taxonomy represented by a hierarchy of resource clusters, using cluster criteria generated from the collected resources. The system includes an editor for editing the criteria, and a renderer for generating linked document data for displaying the hierarchy. A parallel cluster search system is used to evaluate clusters in parallel. The system also includes a parallel classifier for classifying further collected resources.

A comprehensive web based workspace, or search and navigation environment that supports the user throughout an Internet search session and beyond. Specifically, any search session is performed through the workspace which is a web based application that enables the user to log in and identify. The workspace tracks and analyzes the search pattern of the user, learns and characterizes it and provides him or her with ever improving search results, targeted advertisements and other relevant information on-the-fly. Specifically, the workspace is capable of monitoring each and every hypertext document (including URLs) visited by the user. Additionally, the workspace further enables a user to create a cluster search materials comprising URLs and downloaded files. The cluster is used by the workspace to further optimize the search. Moreover—the cluster may be shared between users.

The invention discloses a two-dimensional recursive network-based recognition method of Chinese text in natural scene images. Firstly, a training sample set is acquired, and a neural network formed bysequentially connecting a deep convolutional network, a two-dimensional recursive network used for encoding, a two-dimensional recursive network used for decoding and a CTC model is trained; test samples are input into the trained deep convolutional network, and feature maps of the test samples are acquired; the feature maps of the test samples are input into the trained two-dimensional recursivenetwork, which is used for encoding, to obtain encoding feature maps of the test samples; the encoding feature maps of the test samples are input into the trained two-dimensional recursive network, which is used for decoding, to obtain a probability result of each commonly used Chinese character in each image of the test samples; and clustering searching processing is carried out, and finally, the overall Chinese text in the test samples is recognized. According to the method of the invention, space/time information and context information of the text images are fully utilized, the text imagepre-segmentation problem can be avoided, and recognition accuracy is improved.

The invention discloses a Chinese word segmentation method based on depth learning, comprising the following steps: Chinese characters are maped into character vector based on literal character frequency; the character vector is refined to extract the feature vector with context semantic information and the feature vector with character feature; the character-level vectors are effectively fused with the word-level distributed representation, and then the fused candidate vectors are sent into the depth learning model to calculate the sentence scores, which are decoded by the cluster search method, and finally the appropriate word segmentation results are selected by the sentence scores. In this way, the task of word segmentation can be freed from the tedious feature engineering, better system performance can be obtained by extracting more abundant feature information, and the whole segmentation history can be used for modeling, which has the ability of word segmentation at the sequencelevel.

The invention discloses a Chinese label extraction method for clustering search results of a search engine, which comprises the following steps of: S1, inputting search words by a user to form an input document; S2, selecting candidate words, and scoring all the candidate words; S3, judging whether unmarked candidate words exist, if not existing, skipping to a step S8; if existing, selecting a candidate word with highest score, expanding the selected candidate word into a set of ordered word sequences containing the word, and entering a step S4; S4, calculating the frequency of each ordered word sequence, and extracting the high-frequency word sequence; S5, scoring the high-frequency word sequence, and selecting a candidate word sequence; S6, judging whether the candidate word sequence is accepted as a label, if so, entering a step S7, otherwise, returning to the step S3; S7, performing clustering according to the generated label; and S8, completing the operation. The method can reduce noise labels, and the labels have better representativeness, simplicity and integrity.

A unique device name is assigned to each of a plurality of shared storage devices in a cluster configuration database defining membership of nodes in a cluster. A particular node among the nodes defined by the cluster configuration database as a member of the cluster searches the cluster configuration database for a device identifier matching a device identifier of a shared storage device hosted by the particular node. In response to finding a matching device identifier in the cluster configuration database, the particular node renames, in a local configuration maintained at the particular node, a storage device associated with the matching device identifier with the unique name assigned to that storage device in the cluster configuration database.

The invention discloses an electronic device, a face image cluster search method, and a computer readable storage medium. The method comprises the steps: carrying out the face detection of a real-timeimage, so as to determine a real-time face image and calculate the feature vector of the real-time face image; calculating the distance between the feature vector of the real-time face image and a central point feature vector of each image class determined in advance, screening out the central point feature vector with the distance being less than a first threshold value, and taking the image class corresponding to the screened central point feature vector as a candidate class; searching a sample image from the candidate class, wherein the distance between the feature vector of the sample image and the feature vector of the real-time face image is minimum and is less than a second threshold value. According to the technical scheme of the invention, the method reduces the time in sample comparison in a human face recognition process, and improves the recognition real-time performances of a human face.

The invention provides a client feature library generating method, which comprises the steps of: calculating the sum d of intra-class distances under different cluster numbers k, and drawing a k-d value curve; finding a turning point according to the k-d value curve, finding a k value corresponding to the turning point, looking for an optimal cluster number kopt within a range near the k value, and utilizing a cluster result corresponding to the optimal cluster number as a final cluster result. The invention also provides a client feature library generating device. The scheme of the invention can improve the cluster searching efficiency and precision, and the method of the invention is used for performing client analysis according to the obtained client feature library to improve the recommendation success rate and client satisfaction degree of a marketing campaign.

The embodiment of the invention discloses a data searching method and device. The method comprises the steps that the query per second is calculated by a target server receiving search requests; when the query per second is smaller than an upper threshold value of the query per second, the target server searches for target data meeting the preset requirements in a local database; when the target data is found, the target data is returned to a target client; when the target data is not found, a second server cluster is made to search for the target data in a local cluster database; the second server cluster feeds the found target data back to the target server, and then the target data is returned back to the target client; under the situation that the second server cluster does not find the target data, and the query per second is smaller than a lower threshold value of the query per second, a third server cluster searches for the target data in the local cluster database, and the target data is fed back to the target server; the target server returns the target data to the target client. By means of the data searching method and device, the stability and effectiveness of search service can be improved.

The invention discloses a photovoltaic array multi-peak maximum power cluster searching optimization tracking method, comprising steps of measuring output voltage and output current of a photovoltaic array; according to photovoltaic array component and shadow condition, acquiring peak value number n; using IGSO algorithm and tracking the voltage corresponding to the maximum photovoltaic power; making an impulse generator generate PWM signal according to voltage and then making a photovoltaic array run at an updated array voltage point as the maximum output power of photovoltaic array; when the power change rate is more than 0.015, restarting a searching process. According to the tracking method of photovoltaic array MPPT problem based on IGSO, number of member producer in the IGSO algorithm is arranged according to the array structure, thus the tracking efficiency is improved; according to the rule relationship between the maximum power point and the voltage, the initial position and the searching strategy are improved, and rover is omitted, and thereby avoiding partial extreme point and improving the algorithmic precision and stability.

The present invention discloses a Chinese text overall recognition method in a natural scene image. The method comprises: acquiring a training sample set, and training a neutral network formed by sequentially connecting a deep convolution network, a multi-layer bidirectional recursive network for encoding, a multi-layer bidirectional recursive network for decoding, and a CTC model; inputting a test sample into the trained deep convolution network to obtain each feature sequence frame of the test sample; inputting each feature sequence frame of the test sample into the trained multi-layer bidirectional recursive network for encoding to obtain each code sequence frame of the test sample; inputting each code sequence frame of the test sample to the trained multi-layer bidirectional recursivenetwork for decoding, and obtaining each common Chinese character in each frame image of the test sample as a probability result of the Chinese character in the frame image; and carrying out cluster search processing to finally recognize the overall Chinese text in the test sample. According to the method disclosed by the present invention, the accuracy of the Chinese text overall recognition in the natural scene image is improved.

Systems and methods are disclosed herein for providing search results, particularly product records from a product database. Past queries are analyzed and grouped into clusters according to similarities, with each query including a highly relevant head query and a plurality of tail queries. Similarity of queries may be determined based on user response similarity to query results, co-occurrence, and textual similarity. One or more categories are identified for the clusters, such as based on click-through rates for search results of the head queries of the clusters. Upon receiving a query, a cluster for the query is identified, such as according to similarity to one or more queries of the cluster. The categories associated with the cluster are then used to one or both of augment the query and rank search results for the query.

The invention relates to an artificial intelligence technology, and discloses a keyword generation method, which comprises the following steps of obtaining text data, and identifying the text data by using a preset identifier to obtain a training data set; training by using the training data set to obtain a keyword generation model; receiving a to-be-processed text, extracting semantic information of the to-be-processed text by utilizing the keyword generation model, and generating a semantic vector by adopting an attention mechanism; and utilizing the keyword generation model, based on a preset penalty factor, performing keyword extraction on the semantic vector by adopting a cluster search mode, and outputting the extracted keywords. The invention further provides a keyword generation device and equipment and a computer readable storage medium. In addition, the invention also relates to a blockchain technology, and the text data can be stored in the blockchain node. According to the method and the device, the accuracy of keyword generation can be improved, the repeatability of the generated keywords is reduced, and the coherence between the generated keywords is enhanced.

Typical search engines include document snippets. There is no progression from a big picture view of search results to more detailed views. This invention creates a big picture view of search results in terms of cluster/category/topic/facet summaries/snippets. Cluster summary/snippet also has a few top ranking document snippets unlike other clustering search engines. Interactivity is provided for the user to see more/less results of each cluster/category/topic/facet. This helps in better understanding and navigation of search results. Can use traditional windows/rectangles for clusters/categories and sub-rectangles for documents. Alternatively can use a modified treemap. Each cluster with title becomes a parent in the treemap. Unlike traditional treemaps which show all children at a level, it shows a subset of children. A further alternative uses circles/balloons to represent clusters. Documents are shown as sub-circles. Clustering is also improved by additionally using ‘similar terms’ for clustering and getting cluster titles.

A computing device having an optimized file system and methods for use therein. File system optimizations include sector-aligned writes, anchored cluster searches, anchored index searches, companion caches dedicated to particular file management data types and predictive cache updates, all of which expedite processing on the computing device. The file system optimizations are especially advantageous for data collection systems where an embedded device is tasked with logging to a target memory data received in a continuous data stream and where none of the streamed data is deleted until after the target memory has been offloaded to another device.

The invention provides a DCNN based 3D shape classification method. The method mainly comprises data input, initialization of convolutional neural network, clustering searching and knowledge migration. The convolutional neural network is used, a relatively simple structure of the convolutional neural network serves as a root node of a searching tree, and a cluster searching method is used to explore a candidate more-complex model from the root node; and when a new candidate convolutional neural network is generated, a mother convolutional neural network transmits a proper parameter value to a later generation, a cluster searching result is valid, and an optimal convolutional neural network is obtained finally. Compared with the prior art, performance of a popular 3D shape data set is higher, and the total number of parameters is reduced by about 98%; and knowledge migration is carried out after the cluster searching method, and the method of the invention is can be easily applied to deep robustness learning and the like needed in a mini training data set.

A method and apparatus for clustering search terms are provided by the present invention. The method includes: A, establishing a candidate search term set, wherein the candidate search term set comprises a first search term provided by a user, and a second search term related to the first search term; B, performing a clustering operation on the first search term and the second search term related to the first search term in the candidate search term set according to text characteristic and/or semantic characteristic of search term. The accuracy and relevance of search term clustering can be improved by use of the method.

The invention discloses a k-means algorithm-based public security crime class case research and judgment method. The method comprises the steps of collecting information of solved cases in a time at least greater than a month recently from clients, storing the information in a database, and defining 6 dimension vector attributes; extracting case features of the cases, and performing attribute vectorization by utilizing a bag-of-words model to obtain a case matrix; performing clustering by applying a k-means algorithm to form a class case library, taking a mean value of coordinates of all case vectors in each class set as a centroid Ai of the class set, and forming a vector matrix A by centroids of K classes, wherein i is equal to an optimal value of K; and inputting new cases through users, determining eigenvectors of the corresponding cases through the vector attributes defined in the step 1, namely, inputting distances between the case vectors and k class case sets, and pushing the class case set with the shortest distance to case handling policemen, thereby searching for general characteristics of the cases and assisting in case solving. According to the method, class case clustering search and judgment can be performed intelligently, automatically and accurately, so that the workload of the policemen is greatly reduced and the case solving efficiency is improved.

The invention discloses a method for batch computing of the evolutionary rate of orthologous genes of a genome. According to the method for batch computing of the evolutionary rate of the orthologous genes of the genome, an InParanoid program for clustering searching of orthologous genes after a paired comparing result is obtained based on Blast, an MUSCLE program based on multi-sequence comparative analysis and a PAML software package yn00 program for evaluating the synonymous and non-synonymous substitution ratio based on comparison between coding protein DNA sequences are comprehensively used, and the methods such as Perl scripting language programming are combined. It is proved through experiments that a method comparing system for batch computing of the evolutionary rate of the orthologous genes of the genome is good in repeat effect on various parameter values for detecting the genome and computing the evolutionary rate of the orthologous genes of the genome, high in speed and capable of easily achieving batch computing, automation and processing.