602 results about "Text matching" patented technology

Batch, or “offline”, blocking takes a set of records and generates sets (or blocks, hence the name blocking) of potentially matching records for the entire set. The blocks of potential matches are then passed to a matching process to evaluate which records match. Applications include but are not limited to individual matching such as student identification, householding, business matching, supply chain matching, financial matching, news or text matching, and other applications.

The invention provides a deep text matching method and device based on word migration learning, and the method comprises the steps: firstly, carrying out the fusion of a BERT model and carrying out the pre-training of the BERT model during the training of a deep matching model; secondly, utilizing a pre-trained BERT model to respectively represent sentences in the input sentence pairs by using initial word vectors, and then performing similarity weighting on the sentences in the sentence pairs represented by the initial word vectors to obtain weighted sentence vectors; and finally, according to the loss value corresponding to the similarity value of the statement vector, adjusting model parameters, and carrying out text matching on the input statement by utilizing a depth matching model obtained through parameter adjustment. The parameters of the pre-trained BERT model are no longer randomly initialized parameters, and part-of-speech prediction is added into the pre-trained BERT model,so that the word vector semantic information is enriched. Therefore, semantics, represented by word vectors, of sentences in the sentence pairs are more accurate through the trained BERT model, and the matching accuracy of the trained model is promoted to be improved.

The inventive embodiment discloses an indexing system based on a semantic body. The system comprises a semantic body index database and a semantic body indexing processing unit. The semantic body indexing processing unit is used for acquiring a text hit file list, matching the text hit file list with a semantic body index in the semantic body index database to obtain a file semantic classification list. The indexing system can identify the semantic information of a file to be indexed, and the semantic classification result is presented in the searching result. The invention also discloses an indexing method based on semantic body, which comprises constructing semantic body index for the file with constructed text index, and performing semantic body index matching to the text matching result when user is searching, so that the semantic classification is presented in the final output result based on the conventional text matching result, thereby facilitating user's searching.

The invention provides a method for calling an application program in a mobile terminal, comprising the following steps of: entering an application program quick calling interface; receiving handwriting track information input into the application program quick calling interface by a user and obtaining a corresponding input text by carrying out the handwriting recognition of the handwriting track information; carrying out text matching on the input text and preset application program information; extracting the mapping relation of the matched application program information and a corresponding application program startup instruction; and triggering the corresponding application program startup instruction according to a presetting rule. The invention can simplify user operation and improves the operation efficiency of the application program called by the user, thereby increasing user experience.

The invention discloses a quick content analysis multi-keyword matching method which includes the preprocessing stage and the pattern matching stage; wherein, the preprocessing stage includes the keyword characteristic string clipping, the establishment of the keyword character segmentation set, the Bloom Filter (bloom filter) construction based on the keyword character segmentation set and the original keyword set linear table construction; the pattern matching stage includes the quick judgment that the text string in the current window is not matched with any keyword characteristic segmentation is realized according to the Bloom Filter; the accurate matching of the text string and the candidate keyword is realized through the character string comparison only under the situation that the quick judgment is failed; the text matching window skips in high speed with continuous multibytes. The quick content analysis multi-keyword matching method utilizes the very low success matching rate of the text for being matched and the keyword to realize the high-speed matching under the large quantity of keywords scene so that the multi-keyword matching method is very applicable to the online virus scanning application such as the virus detection.

Personalized messages can be generated according to contents detected in communications between parties to a commercial transaction. The information streams between parties is monitored; spoken keywords, images, video or textual data are detected by pattern recognition, voice recognition or text matching. One or more subjects of the communication detected from recognized keywords, images, video or text data are used to generate a personalized message to one or more participants according to the detected subjects.

The invention relates to the artificial intelligence technology field and particularly relates to an image text matching model training method, a bidirectional search method and a related device. Thetraining method comprises steps that global and local representations of an image sample and a text sample are extracted, a pre-constructed matching model is trained, the global and local representations of the image sample and the text sample are respectively mapped to the specified semantic space by the matching model, and similarity of the global representation and similarity of the local representation are calculated; according to the default weight of the similarity of the global representation and the default weight of the similarity of the local representation, the image and text matching degree is determined through employing a weighted summation mode. The training method is advantaged in that the acquired matching degree is based on considering both detail characteristics of an image and global characteristics, and the acquired matching degree is more accurate and comprehensive.

The invention discloses a page keyword acquisition method which acquires the entry of the key text by word-segmentation to the acquired key text in WEB pages and defines the entry of the key text matching with the predefined entry in the predefined word library as the key words of the WEB page. The invention further discloses a page information processing device. With the invention which is simple and convenient, all the main elements of the WEB pages can be considered, thus improving the accuracy of acquiring the WEB page keywords.

The invention belongs to the technical field of intelligent education question-answer, and particularly relates to a mixed automatic question-answer method based on education knowledge graphs and texts. The mixed automatic question-answer method comprises the following steps: constructing a basic education knowledge graph by constructing a basic education body, semantic annotation and informationextraction; constructing a general template of the question according to the keywords in combination with a regular expression; establishing a full-text search engine, and preprocessing mass texts; training test question and answer pairs as a training set until a deep text matching model converges; identifying the user questions to obtain a subject list, and endowing the subject list with confidence; carrying out template matching to obtain a predicate list, and giving confidence to the predicate list; inquiring a knowledge graph according to the subject and predicate lists to obtain an answerlist, and endowing confidence coefficients; obtaining keywords by using a part-of-speech tagging method, performing coarse-fine granularity matching to obtain answers, and sorting the answers; if thehighest confidence of the answer based on the educational knowledge graph exceeds a threshold value, returning the answer; or, returning the answer with the highest sorting based on the text.

A system and method for matching data using probabilistic modeling techniques is provided. The system includes a computer system and a data matching model / engine. The present invention precisely and automatically matches and identifies entities from approximately matching short string text (e.g., company names, product names, addresses, etc.) by pre-processing datasets using a near-exact matching model and a fingerprint matching model, and then applying a fuzzy text matching model. More specifically, the fuzzy text matching model applies an Inverse Document Frequency function to a simple data entry model and combines this with one or more unintentional error metrics / measures and / or intentional spelling variation metrics / measures through a probabilistic model. The system can be autonomous and robust, and allow for variations and errors in text, while appropriately penalizing the similarity score, thus allowing dataset linking through text columns.

The invention provides a text semantic matching method. The method comprises the steps of obtaining total vectors by using word vectors and character vectors corresponding to words and characters in ato-be-processed text; performing convolution processing, averaging processing and bidirectional LSTM processing on the total vectors, wherein an LSTM method without an attention mechanism is adoptedduring basic text processing, and an LSTM method with the attention mechanism is adopted during target text processing; obtaining pooling vectors; and taking the similarity of the two pooling vectorsas the similarity of two texts. According to the method, not only a semantic relationship among the words in the text but also a semantic relationship among the characters are considered; not only a positive semantic relationship between the characters and the words but also a negative semantic relationship between the characters and the words are considered; and multiple text matching models arecomprehensively applied, so that feature information among text contexts is expressed more comprehensively, accurately and fully, and the text matching accuracy is improved.

The embodiment of the invention provides a text matching method and device applied to electronic equipment. The method comprises the steps that search terms input by a user are received and subjected to participle treatment, at least one participle is obtained; according to a pre-trained classification model and at least one of the term property, semantic attribute, independent search term probability and click rate of each term, the importance weight of each participle is determined; according to the text to be matched and a text set where the text to be matched is located, the term frequency and inverse document frequency of each participle are determined; according to the importance weight, term frequency and inverse document frequency of each participle, the matching degree of the search terms and the text to be matched is calculated. According to the text matching method and device, the importance degree of each participle can be more accurately measured by means of the important weight and inverse document frequency of each term, and therefore the text matching accuracy can be improved when text matching is carried out according to the importance degree of each participle.

An image search method and system using face detection, begins with receiving a query submitted by a user. Next a query word is searched in the query from an image resource using an image search engine to obtain an initial image collection. Any faces are detected in each image in the initial image collection which has been searched. A search for the query word in a text surrounding each image having the face in the initial image collection is performed. A determination is made whether the query word indicates at least one person's name in the surrounding text matching the query word. An image an image in the initial image collection is returned to a user in which the face is included and the query word in the surrounding text indicates the person's name.

Methods and apparatuses for automatic treatment planning, including recommendation systems, quality assurance, error prevention, text mining, text matching, and treatment planning optimization.

The invention relates to a Chinese variation text matching recognition method. The method comprises the following steps of: performing special encoding conversion on a target text and a mode character string to improve the similarity of texts; adding proper wildcard characters into the converted mode character string according to the structural characteristics of Chinese characters in the mode character string; and matching by using a precise character string matching algorithm (namely a CV-BM algorithm). The method solves the problems of the replacement of characters with similar shapes in the variation text and the matching difficulty based on the replacement of wrongly written characters with similar shapes. The problems of the replacement of the characters with the similar shapes in the variation text and the matching difficulty based on the replacement of wrongly written characters with the similar shapes can be solved. Compared with an image partitioning recognition method, the method has smaller time and space complexities, is more suitable for the quick matching of Chinese character information in a high-speed network data transmission environment, and can be widely applied to systems needing to match the Chinese character information, such as an intrusion prevention system, an information retrieval system and the like.

The invention discloses a method, a device and computer equipment, which are used for identifying short text category information and are implemented by a computer. According to the technical scheme provided by the invention, the method comprises the following steps of: acquiring the short text category information to be classified; acquiring a plurality of category judged results of the short text information, which are acquired on the basis of a plurality of pre-trained classification model groups respectively; and on the basis of a first predetermined rule and according to the category judged results, comprehensively determining the category information of the short text information to be classified. Compared with the prior art, the invention has the following advantages that: 1) the category of the short text information is comprehensively judged through characteristic information of a plurality of short texts, so that the judgment accuracy is improved; 2) the automatic expansion of classified short text information is realized through iterative training and classification; and 3) the category information of the short text information can be acquired in real time by generating a short text matching dictionary.

Provided is a method, system, and program for encoding and decoding input data. A key is generated comprising a list of a plurality of binary values, wherein the key is available to a user during decoding. The input data is encoded by using the key to initialize a replacement list. A determination is made whether to replace strings in the input data with a string reference to a matching string value, wherein the input data not replaced with reference to one matching string value comprises one or more literals. Each literal is replaced with a literal reference to one entry in the replacement list matching the literal. The encoded input data is decoded by accessing the generated key in response to user input and using the accessed key to decode the literals replaced with literal references to entries in the replacement list.

The embodiment of the invention discloses an image-text matching method and device, a storage medium and equipment, and belongs to the technical field of computers. The image-text matching method comprises the steps of obtaining an image and a text to be matched; generating a candidate instance feature set according to the image; aggregating candidate instance features in the candidate instance feature set by using a self-attention mechanism to obtain an instance feature set, each instance feature in the instance feature set corresponding to one object in the image; encoding the text to obtaina text vector; and calculating a matching degree between the image and the text according to the instance feature set and the text vector. According to the embodiment of the invention, the realization difficulty of image-text matching can be simplified, and the accuracy of image-text matching can be improved.

The invention discloses a cross-modal retrieval method based on multilayer semantic alignment, which comprises the following steps of: acquiring a remarkable fine-grained region by utilizing a self-attention mechanism, promoting the alignment of entities and relationships among modal data, providing an image text matching strategy based on semantic consistency, extracting semantic tags from a given text data set, and performing global semantic constraint through multi-label prediction, so that more accurate cross-modal association is obtained. Therefore, the problem of cross-modal data semantic gap is solved.

The invention relates to a method and device for evaluating the level of spoken English. The method comprises the steps of receiving a speech signal inputted by tested person after outputting a test question; performing speech recognition on the speech signal to obtain corresponding text information; performing text matching on the text information with an answer in an answer set corresponding tothe test question, wherein the answer set comprises a standard answer of the test question and derivative answers extended according to the standard answer; obtaining an evaluation result at least according to the text matching result; and outputting the evaluation result. Therefore, effective evaluation can be performed on the meaning and logicality of a spoken language text, the comprehensiveness and accuracy of the evaluation are improved, and the evaluation result is high in feedback speed. In addition, the tested person can obtain corresponding improvement suggestions, thereby being highin pertinence.

The present invention provides a remote sensing image semantic generation method based on a fast region convolutional neural network. The problems are overcome that relation between images in an imagecannot be obtained and a relation between a target and the whole image cannot be obtained. The method comprises the steps of: constructing a training sample set and a test sample set; employing a fast region convolutional neural network to extract image features of a high-resolution remote sensing image; employing a bidirectional recurrent neural network to extract text features of correspondingsentences; employing an image-text matching model based on probability to match the image features with the text features; and employing the matching image-text features to perform training for a long-and-short time memory network for training so as to achieve semantic generation of the high-resolution remote sensing image. The remote sensing image semantic generation method fully considers the features of complex remote sensing image backgrounds and multiple targets so as to improve the result of remote sensing image semantic generation, and can be used for image retrieval or scene classification.

The invention relates to the technical field of natural language processing, and provides a text matching method and device based on transfer learning, a computer storage medium and electronic equipment, and the method comprises the steps: obtaining a query statement and a to-be-matched statement set corresponding to the query statement; on the basis of a pre-trained language model, determining anith semantic vector set about the ith to-be-matched statement and the query statement in the to-be-matched statement set, wherein each semantic vector set comprises a first semantic vector and a second semantic vector; determining a matching value between the ith to-be-matched statement and the query statement according to the first semantic vector and the second semantic vector in the ith groupof semantic vector set, where i is a positive integer; and according to the matching value, determining a target matching text of the query statement in the to-be-matched statement set. According to the technical scheme provided by the invention, the text matching accuracy based on semantic representation can be remarkably improved.

The invention provides a rapid fuzzy matching algorithm for strings. According to the invention, firstly data preprocessing is performed on texts in a database to obtain a statistical model and an index is established via Hash. An input text is a shorter string. The algorithm traverses all Chinese characters therein, activates the positions of corresponding Chinese characters in a finite character complete set, and maps the activation state of the finite character complete set to each tag to filter tags. A few filtered tags are used for matching the texts and the DTW algorithm is used for approximate string matching. The algorithm also comprises the steps of performing scoring and sorting according to the result of the degree of approximation of matching and returning to a search result. Through the efficient tag filtering method, the calculation efficiency of the string matching algorithm is greatly increased; in a process of input text matching, a fuzzy matching effect is achieved and a good matching performance is guaranteed for fuzzy languages.

Systems and methods for speech processing in telecommunication networks are described. In some embodiments, a method may include receiving speech transmitted over a network, causing the speech to be converted to text, and identifying the speech as predetermined speech in response to the text matching a stored text associated with the predetermined speech. The stored text may have been obtained, for example, by subjecting the predetermined speech to a network impairment condition. The method may further include identifying terms within the text that match terms within the stored text (e.g., despite not being identical to each other), calculating a score between the text and the stored text, and determining that the text matches the stored text in response to the score meeting a threshold value. In some cases, the method may also identify one of a plurality of speeches based on a selected one of a plurality of stored texts.

The present application relates to the field of natural language processing technology and provides a contract intention judgement method and apparatus, a computer apparatus, and a storage medium. Themethod comprises the following steps of: acquiring voice data from a terminal; obtaining a speech text by performing speech recognition of speech data, extracting keywords from speech text, matchingthe extracted keywords with preset keywords corresponding to preset text, and obtaining text matching score; extracting the speech feature vector from the speech data, and inputting the extracted speech feature vector into the trained fraud risk prediction model to obtain the fraud risk coefficient value; acquiring image data from a terminal, performing face recognition on the acquired image datato obtain a face recognition score; acquiring scene identifiers corresponding to the current business scenario, and acquiring corresponding target audit rules according to the scene identifiers; according to the text matching score, fraud risk coefficient value, face recognition score and target audit rule, judging the intent to sign the contract.

The invention discloses a method and a device of searching a web browsing history. The method of searching the web browsing history comprises the following steps of receiving a search text which is input by a user; obtaining a pre-generated web browsing history record; judging whether the record is matched with the search text according to search keyword information carried in each record; and generating a search result by utilizing the record clause with the successful matching. In comparison with the prior art, the scheme of the invention is based on a text matching manner, the search effect is remarkably better than the scheme of searching according to a browsing period. In addition, the matching basis is the search keyword, which is more sensitive to the user, so that the search keyword is easier to remember than contents such as a web URL or a web title, and the corresponding search result conforms to the browsing requirements of the user well.

The invention relates to a search sorting method and device, computer equipment and a storage medium. The method comprises the steps of obtaining a search statement and corresponding to-be-sorted recall statements; vectorizing the search statement and each to-be-sorted recall statement to obtain a search statement vector and each to-be-sorted recall statement vector, and calculating the similaritybetween the search statement vector and each to-be-sorted recall statement vector to obtain each similar feature; performing text matching based on the search statement and each to-be-sorted recall statement to obtain each text matching feature; calculating each self-attention matrix corresponding to the search statement and each to-be-sorted recall statement, and performing calculation based onthe self-attention matrixes to obtain each interaction feature; based on the text matching features, the similarity features and the interaction features, calculating the correlation degree between the search statement and each to-be-sorted recall statement; and obtaining a sorting result of each to-be-sorted recall statement according to the correlation degree. By adopting the method, the accuracy of the sorting result can be improved.

The invention provides a text similarity calculation method and system, a data query system, a computer product and a computer readable storage medium. The text similarity calculation method comprisesthe steps that at least a first text and a second text are acquired; the first text and the second text are mapped into vectors; similar parts and different parts of the first text and the second text are calculated; and the similar parts and the different parts are utilized to calculate the similarity between the first text and the second text. According to the text similarity calculation methodand system, the influence of word semantic similarity and difference on the text similarity is at least partially considered, the semantic similarity between the texts can be calculated from finer granularity, therefore, text matching precision is improved, and high-precision retrieval or query and the like are realized.

The invention discloses a self-adaptive intelligent banner advertisement picture generation method. The method is characterized in the experience knowledge of a designer is external, data informationof design principles such as a picture color matching principle, an image-text matching rule, a visual combination structure and the like are extracted; the functions of automatic material matching and picture synthesis of the banner advertisement picture are realized; when the banner advertisement picture is generated, only a to-be-synthesized product picture needs to be input, copywriting information such as main and subtitles is automatically extracted according to the product picture, the best matched background and modification elements are selected from the material library according topicture color matching, finally, the final synthesis format is determined according to the selected elements, and generation of the banner advertisement picture is completed. According to the method,the low-efficiency repeated labor of a designer is effectively reduced, and the cost of synthesizing the banner advertisement is reduced; when the size, the elements and the like of the banner advertisement are changed, the required banner advertisement picture can be dynamically generated in real time, and possibility is provided for individuation.

The invention provides an image-text matching method and device based on artificial intelligence, an electronic device and a storage medium, and the method comprises the steps: obtaining the semanticword features of all words in a given text, and synthesizing the global features of the given text based on the word features; acquiring regional position features of the candidate image for representing a position relationship in the region; acquiring regional visual features of each region in the candidate image; combining the regional visual features and the regional position features to form regional features corresponding to each region of the candidate image, and obtaining global features corresponding to the candidate image; and determining similarity scores of the candidate images andthe given text based on the word features, the global features of the given text, the regional features and the global features of the candidate images, and determining that the candidate images are matched with the given text when the similarity scores are greater than a similarity score threshold. Through the method, image-text matching accuracy can be improved.