121results about How to "Accurate similarity" patented technology

The invention provides a content aggregation method based on distributed web crawlers, which comprises the steps that firstly different crawler platforms are arranged at different devices, a request is sent to a crawling network information source end, and the crawler platforms fabricate crawling rules according to target information required by a user and crawl information in which the target user is interested; the crawled network information is processed, similarity detection is carried out based on a data transmission and conversion method in a real-time database and by being combined with a locality sensitive hashing (LSH) method so as to reduce the redundancy of the information; and the information is classified and sorted by the system according to the category, the heat and keywords and then displayed on user equipment. According to the method provided by the invention, LSH and similarity comparison are carried out according to the data information acquired in an actual network so as to acquire a comparison result. Compared with a comparison result acquired by adopting a traditional mode of whole data duplication checking in the prior art, the content aggregation method is higher in calculation speed and more accurate in similarity comparison.

The invention relates to a cross-media retrieval method based on deep learning and consistent expression spatial learning. By starting with two methods including feature selection and the similarity estimation of two highly-isomerous feature spaces, the invention puts forward the cross-media retrieval method capable of improving multimedia retrieval accuracy to a large extent by aiming at the cross-media information of two modalities including an images and a text. The method disclosed by the invention is a multimedia information mutual retrieval method which aims at two modalities including the image and the text, and cross-media retrieval accuracy is improved to a large extent. In a model which is put forward by the invention, a regulated vector inner product is adopted as a similarity metric algorithm, the directions of the feature vectors of two different modalities are considered, the influence of an index dimension is eliminated after centralization is carried out, an average value of elements is subtracted from each element in the vectors, the correlation of the two vectors subjected to average value removal is calculated, and accurate similarity can be obtained through calculation.

The invention discloses a video keyframe extraction method, which comprises the steps of: using a ViBe algorithm fused with an inter-frame difference method to perform moving object detection on an acquired original video sequence, so as to obtain a key video sequence containing a moving object; performing keyframe crude extraction on the key video sequence by using a global characteristic peak signal-to-noise ratio to obtain candidate keyframe sequences; and establishing global similarity of the candidate keyframe sequences by using the peak signal-to-noise ratio, establishing local similarity of the candidate keyframe sequences by using SURF feature points, and performing weighted fusion on the global similarity and the local similarity to obtain comprehensive similarity, performing self-adoptive keyframe extraction on the candidate keyframe sequences by using the comprehensive similarity, and finally acquiring a target keyframe sequence. The video keyframe extraction method providedby the invention can effectively extract the video keyframes, obviously reduce the redundant information of video data, and express the main content of the video concisely. Moreover, the video keyframe extraction method has low algorithm complexity and is suitable for real-time extraction of keyframes of surveillance videos.

The invention discloses an individualized recommendation method based on a knowledge map, and belongs to the technical field of knowledge maps and machine learning. The method comprises the followingsteps: 1, carrying out vectorization on goods in the knowledge map so as to acquire a vector set D and quantized values of each goods; 2, calculating goods semantic similarity between the objects according to the quantized values acquired based on the knowledge map; 3, calculating goods interaction similarity between the goods in user historical interaction data based on user behaviors; 4, calculating goods fusion similarity of the all goods according to the goods semantic similarity and the goods interaction similarity; and 5, scoring goods which are not evaluated according to the goods fusion similarity, and generating recommended lists for users according to the scores. According to the individualized recommendation method based on the knowledge map disclosed by the invention, through combination of the goods semantic similarity based on the knowledge map and the goods fusion similarity based on the user behaviors, the recommendation effect of a recommendation system is improved.

An image retrieving system is provided which is suitable for acquiring a retrieving result or a classifying result according to the desire of a user. The system extracts a noticing area from a retrieving key image and each retrieving object image, and a characteristic vector V of these images is generated on the basis of the extracted noticing area. An image similar to the retrieving key image is retrieved from a retrieving object image registration database on the basis of the generated characteristic vector V.

A method is disclosed for translation of an input sentence in a source language to an output sentence in a target language using a store comprising a plurality of example sentences in the source language each paired with its translation in the target language. The method provides for improved matching of the input text against the store of example sentences by analysing both the sentences in the store and the input sentence using a bilingual resource combining aspects of a bilingual dictionary and thesaurus in order to determine the senses and translations of the words in the input sentence and the examples.

In an index generating device, a similarity calculating unit calculates access similarities indicating similarities of access histories between documents, based on history information indicating the access histories to the respective documents by users, and a similar document specifying unit specifies a similar document similar to a given document as a retrieval target, based on the access similarities or the like. A retrieval index generating unit generates a retrieval index for the given document as the retrieval target from words appearing in a document set consisting of the similar document and the given document as the retrieval target.

The invention discloses a collaborative filtering method on the basis of a scene implicit relation among articles. The collaborative filtering method comprises the following steps of: 1, extracting scores of the articles in different scenes from original score data and establishing an article-scene score matrix; 2, decomposing the article-scene score matrix by a matrix decomposition method to obtain an implicit factor matrix of the articles; 3, establishing a scene feature vector for each article by using the obtained implicit factor matrix of the articles so as to calculate the similarity among the articles by utilizing a Pearson correlation coefficient and establish an article implicit relation matrix; and 4, integrating obtained article implicit relation information into a probability matrix decomposition matrix to generate a personalized recommendation. According to the invention, scene information can be sufficiently utilized to mine the implicit relation information among the articles, and the recommendation is generated by utilizing the implicit relation among the articles; the collaborative filtering method has high expandability for the scene information, and a candidate scene set can be regulated according to the application requirements; and the accuracy and the personalization degree of the recommendation can be effectively improved.

The invention provides a heterogeneous media similarity calculation method and heterogeneous media retrieval method based on correlation analysis. The heterogeneous media similarity calculation method includes the following steps that a heterogeneous media database including different media types is established, and the feature vector of data of each media type is extracted; based on the incidence relation inside media, the heterogeneous media similarity can be calculated through K nearest neighbor analysis; based on the incidence relation between media, the heterogeneous media similarity can be calculated through heterogeneous media constraint transmission; the content similarity inside the media and between the media can be fused through a self-adaptation sorting result fusion algorithm so that a final result can be obtained, the fusion weight is set in a self-adaptation mode, and accordingly the final heterogeneous media similarity can be obtained to be used for heterogeneous media retrieval. According to the heterogeneous media similarity calculation method and retrieval method based on correlation analysis, the class information inside the media and the constraint information between the media are fully considered, different similarity calculation methods can be fused in a self-adaptation mode, different media can be mutually promoted, the similarity calculation accuracy is improved, and therefore higher heterogeneous media retrieval accuracy can be achieved.

The invention discloses an online learning-based potential semantic cross-media hash retrieval method, which realizes cross-media retrieval of image and text modes. The method comprises the followingsteps of establishing an image and text pair data set, extracting features of data in the data set, performing mean removal, and dividing the data set into a training set and a test set according to acertain ratio; mapping discrete tags to continuous potential semantic spaces, and building an objective function by utilizing the similarity between the retention data; solving the objective functionby utilizing an online learning-based iterative optimization scheme, and when new data is generated, updating a hash function by only utilizing the new data, thereby improving the efficiency of a training process; and calculating hash codes of the image and text data in the test set by utilizing the hash function, by taking the data in one mode in the test set as a query set and the data in the other mode as a target data set, calculating Hamming distances between the data in the data query set and all the data in the target data set, performing sorting according to an ascending order, and returning the heterogeneous data sorted in front to serve as cross-media retrieval results.

The invention discloses an image classifying method based on transfer learning a multiple attractor cellular automata (MACA) and mainly solves the problems that an existing image classifying method based on transferring cannot avoid the empty basin appearance, the similarities of a source domain sample and a target domain sample are not accurately calculated, the effect of transferring from the source domain sample to the target domain sample is poor, and the classification accuracy is low. The image classifying method comprises the steps: (1) performing image data pre-processing; (2) training a multiple attractor cellular automata (MACA) tree in a source domain space; (3) dividing a target domain training set; (4) constructing a local mode space training set; (5) training a multiple attractor cellular automata (MACA) tree in a local mode space; (6) generating a target domain multiple attractor cellular automata (MACA) tree. The image classifying method has the advantages of strong generalization ability and high classifying accuracy and effectively solves the problems that the existing image classifying method cannot avoid the empty basin appearance and poor transferring effect.

The embodiment of the invention discloses a similar anchor classification model training method, an anchor recommendation method and related devices. The training method comprises the steps of obtaining historical behavior data of watching live broadcast by a user; determining a plurality of first candidate anchor pairs according to the historical behavior data, wherein each first candidate anchorpair comprises two anchors; obtaining anchor information of each anchor; for each first candidate anchor pair, anchor pair features of the first candidate anchor pair are extracted based on historical behavior data and anchor information; determining an anchor pair sample from the plurality of first candidate anchor pairs, wherein the anchor pair sample comprises an anchor pair feature and an anchor pair label; and obtaining a similar anchor classification model for outputting the anchor similarity by adopting the anchor pair features of the anchor pair samples and the anchor pair label training model. According to the embodiment of the invention, the similarity of the two anchors can be determined from multiple dimensions and the trained similar anchor classification model, the universality is good, and the similarity of the anchors is accurate, so that the anchor recommendation accuracy is improved.

The invention provides a tourist portrait construction method for scenic spot recommendation. The method comprises steps of generating a feature vector of the tourist history sightseeing scenic spot sequence relative to the candidate scenic spot by utilizing the propagation of the tourist history sightseeing scenic spot sequence on the tourist knowledge map; distributing different weights to different feature vectors through the attention network; calculating to obtain a weighted sum of the feature vectors; wherein the weighted sum is a representation vector of the tourist, taking the obtained representation vector as a representation of a tourist portrait, carrying out inner product operation on the scenic spot representation vector and the tourist portrait at a personalized scenic spot recommendation stage to generate a scenic spot touring probability of the tourist, and generating a scenic spot recommendation list for the tourist according to the probabilities of different scenic spots.

The invention provides a similar model retrieval implementing method based on three-dimensional labeling. By means of the method, the function of extracting three-dimensional model labeling information is achieved through three-dimensional CAD software, and a three-dimensional model labeling feature database is formed with extracted size labeling information. The size labeling information of the current three-dimensional model is extracted when similar retrieval needs to be conducted on the three-dimensional model, and the extracted size information is processed, the similarity between the size information and the model size classification and counting standard value is calculated, and the similarity of the current three-dimensional model is obtained. Then, one or more pieces of three-dimensional model data the highest in the current model similarity are retrieved and returned according to the calculated similarity, and a similar data set is formed. The similarity between the size information and the size information of each three-dimensional model in the similar data set is calculated secondarily, similarity sequencing is conducted again, and in combination with a light-weight model, the result is presented in a visual mode. By means of the method, work efficiency is remarkably improved, and design capacity is improved.

An image retrieving system is provided which is suitable for acquiring a retrieving result or a classifying result according to the desire of a user. The system extracts a noticing area from a retrieving key image and each retrieving object image, and a characteristic vector V of these images is generated on the basis of the extracted noticing area. An image similar to the retrieving key image is retrieved from a retrieving object image registration database on the basis of the generated characteristic vector V.

The embodiment of the invention discloses a video recall method and device and a storage medium, and is applied to the technical field of information processing. The video recall device extracts the feature information of the multi-modal data of each video in the video pair, fuses the extracted feature information to obtain the comprehensive feature information of each video, calculates the similarity between the videos in the video pair according to the comprehensive feature information, and then recommends the videos according to the calculated similarity. According to the embodiment of theinvention, in the process of calculating the video similarity, the feature information of the multi-modal number is adopted based on the features of the video content, so that the video recall devicecan calculate the similarity according to the content of the video instead of according to the watching behavior parameters of the video, and the loss of the video feature information is avoided; in addition, due to the fact that the characteristics of the multiple types of data based on the video content are adopted in the embodiment, one video can be described comprehensively, and the calculatedsimilarity is more accurate.

A similarity analyzing device includes: an image acquisition section which acquires picked-up images with which image pick-up dates and / or times are associated; and an image registration section which registers a face image showing a picked-up face and with which an image pick-up date and / or time is associated. The device further includes: a degree of similarity calculation section which detects a face in each of picked-up images acquired by the image acquisition section and calculates the degree of similarity between the detected face and the face in the face image registered in the image registration section; and a degree of similarity reduction section in which the larger the difference between the image pick-up date and / or time associated with the picked-up image and that associated with the face image is, the more the degree of similarity of the face calculated by the degree of similarity calculation section is reduced.

The invention discloses a method for determining document similarity based on improved Jaccard coefficients. The method comprises: step 1, respectively determining the corresponding number (which is shown in the description) of each element wi, the length of which is K in a document X, and the corresponding number (which is shown in the description) of each element wj, the length of which is K in a document Y; step 2, calculating the proportion (which is shown in the description) of each element wi in the document X; step 3, calculating the proportion (which is shown in the description) of each element wj in the document Y; step 4, calculating the Jaccard similarity (which is shown in the description) of a common element wh in the document X and the document Y; step 5, calculating the Epsilon (wh) of the element wh in elements, the n-Gram length of which is K in the document X and the document Y; step 6, calculating a parameter F(wh) representing whether the element wh simultaneously exists in the document X and the document Y; and step 7, setting a symbol (which is shown in the description) as the similarity of the document X and the document Y. According to the method for determining the document similarity based on the improved Jaccard coefficients, by considering the proportion of each element and sample in the documents and the contribution degree to the similarity of multiple documents, a problem that inter-document similarity calculation is inexact in the prior art is effectively solved.

The embodiment of the invention discloses a data processing method and device. The method comprises the following steps of acquiring the historical behavior data of all users and generating a user behavior topological network according to the historical behavior data of all users; wherein the user behavior topology network comprises a plurality of sub-behavior topology networks, and each sub-behavior topology network comprises at least a first user node and a second user node; generating a first splicing vector corresponding to the first user node and generating a second splicing vector corresponding to the second user node according to the sub-behavioral topological network; acquiring a vector distance value between the first splicing vector and the second splicing vector, and determiningthe user similarity between the first user node and the second user node according to the vector distance value. By adopting the invention, the similarity between any two nodes can be accurately andreasonably quantified, and the calculation error can be reduced.

The invention discloses an XML (Extensive Makeup Language) structural similarity measuring method based on a frequency-associated tag sequence. The method comprises the following steps of: resolving an XML document set C to obtain a tag sequence database (TSDB); excavating all frequency tag sequence sets (FTS) from the TSDB; selecting a maximum frequency tag sequence set (MFTS) from the FTS; converting to obtain a new TSDB'; excavating a closed frequency-associated tag sequence set from the TSDB'; and expressing any document in the TSDB' as a closed frequency-associated tag sequence set whichis contained in the TSDB', and calculating the structural similarity between any two documents in the document set C. According to the method, the accuracy of a clustering result can be raised.

The invention discloses a target account determination method and device, a storage medium and an electronic device. The method comprises the following steps of: training a target model according to the obtained first account feature vector of the seed account; wherein in the training process, the feature vectors serving as the input accounts input to the target model comprise the first account feature vectors of the seed accounts, and the feature vectors of the input accounts are vectors expressed at least by using the portrait feature vectors and the interaction feature vectors; determiningthe similarity between the to-be-determined account and the seed account through the trained target model, and in the determining process, the feature vector serving as the input account input to thetarget model comprising the second account feature vector of the to-be-determined account; and determining a target account of which the similarity meets a target similarity condition in the to-be-determined accounts. According to the method, the technical problem of low determination efficiency of the target account in the prior art is solved.

The invention discloses a production equipment for bamboo mats and bamboo slices, which comprises a box body, one side of the inner wall of the box is fixedly connected with a support plate, and the top of the support plate is fixedly connected with a fixed plate, and one side of the fixed plate is fixed The first motor is connected, and the output shaft of the first motor is fixedly connected with a cam, the top of the cam is connected with a moving rod, one end of the moving rod is slidingly connected with the fixed plate, and one side of the fixed plate is provided with a A slide rail matched with a moving rod, one end of the moving rod is fixedly connected with a second motor, and the invention relates to the technical field of bamboo product production equipment. The bamboo sheet production equipment for bamboo mats can ensure the stability of the hole position, and the cutting and drilling are done in one go, and the production efficiency is high. After the bamboo sheet is punched, the shape can be polished based on the hole position, so that each The similarity of the bamboo slices is more accurate, and the processed bamboo slices are smooth and complete, which ensures the quality of the mat.

The invention provides a method and apparatus for measuring a similarity of documents. The method comprises the steps of acquiring a target document and at least one comparison document; carrying out word segmentation on the target document and the at least one comparison document to acquire words to be processed of each comparison document and words to be processed of the target document; according to an occurrence frequency of each word to be processed in each comparison document, generating a comparison semantic vector of each comparison document, and according to an occurrence frequency of each word to be processed of the target document, generating a target semantic vector of the target document; and according to the target semantic vector and each comparison semantic vector, determining the similarity of the target document and each comparison document. According to the method and apparatus for measuring the similarity of the documents, which are provided by the invention, the similarity between the documents can be more accurately determined.

The invention discloses an industrial system fault detection method based on Euclidean distance multiscale fuzzy sample entropy. According to the method provided by the invention, the complexity of atime sequence can be described from a plurality of time scales; meanwhile, compared with an existing multiscale entropy method and an existing composite multiscale entropy (FME) method, the method hasthe advantages that the stability and the accuracy of the calculation of multi-scale fuzzy sample entropy (FME) are remarkably improved. The method can be used for judging and detecting the fault type of an industrial system and analyzing the complexity of the time sequence.

A first image and a second image obtained by imaging the same subject with different types of modalities are obtained. The first image is deformed, and similarity between the deformed first image and the second image is evaluated by an evaluation function that evaluates correlation between distributions of corresponding pixel values of the two images to estimate an image deformation amount of the first image. Based on the estimated image deformation amount, a deformed image of the first image is generated. The evaluation function includes a term representing a measure of correlation between a pixel value of the deformed first image and a corresponding pixel value of the second image, wherein the term evaluates the correlation based on probability information that indicates a probability of each combination of corresponding pixel values of the first image and the second image.

The invention relates to an inter-text similarity calculation method and device, a storage medium and electronic equipment. The method comprises the steps that word segmentation and stop word filtering processing is conducted on a first text and a second text with the similarity to be calculated, and a first word segmentation set which does not contain repeated word segmentation and corresponds tothe first text and a second word segmentation set which corresponds to the second text are obtained according to the processing result; Determining the semantic information transfer cost between thefirst text and the second text according to the information amount of each segmented word in the first segmented word set and the second segmented word set in the text and the word embedding vector corresponding to each segmented word; And determining the similarity between the first text and the second text according to the semantic information transfer cost. Therefore, the semantic influence ofeach word in the text and the context of each word on the text is fully considered, and the calculation basis of the similarity is closer to the semantics of the text, so that the calculated similarity is more accurate.