139 results about "Word identification" patented technology

The invention discloses a voice wake-up method and apparatus, a terminal, and a processing method thereof. The voice wake-up method comprises: a wake-up word voice signal sent out by a terminal user is collected and a wake-up word identification result and a vocal print feature of the wake-up word voice signal are obtained and stored; when a terminal is in a sleep state, a voice signal inputted externally is detected; a voice identification result and a vocal print feature of the voice signal are obtained; the voice identification result and the stored wake-up word identification result as well as a matching result of the vocal print feature and the stored vocal print feature are obtained; and according to the matching result, whether the terminal needs to be woken up is determined; and if so, the terminal is woken up. With the method, technical problems of low user wake-up rate, high wrong wake-up rate, and poor user experience according to the existing voice wake-up technology are solved.

A image processor of automatic calibrating aberration makes use of radon transforming to pick up four sides profiles of business cards and texts, with profile vertex message, we can calibrate the image aberration and establish internal mathematics module of shot equipment and equation groups of business cards and texts aspect ratio parameter, then solve equation groups to get the rectangle object's aspect ratio and apply it to calibrate images to prevent distortion. At last, the invention can identify the image inversion by words identification for judging if it is necessary to circumvolve the calibrated images to get the perfect calibration results. It is very easy to calibrate images without manual practice , we can utilize the invention in most of the imaging equipments ,such as medical imaging, supervising equipment, mobile, digital camera ,digital vision and so on, no wonder that every user can do the operation very easily.

The invention relates to an automatic webpage classification method based on network hot word identification. The automatic webpage classification method mainly comprises the following steps of: acquiring webpage content information by using customization crawlers, and automatically performing word classification on acquired webpage contents through an Internet keyword base and an Internet stopword base; calculating a hot value according to a keyword appearance frequency and a time distance degree, and performing initial classification on the webpage contents according to the hot value of the words through a Bayesian multidimensional classification model; performing relevance identification on non-matched word classification items in classified webpages through a relevance algorithm, and finding out non-collected hot words from the Internet keyword base, and collecting the non-collected hot words into the Internet keyword base; and reclassifying the webpage contents which cannot be classified in an initial webpage classification process through an updated Internet word base.

The invention relates to the technical field of biomedicine, and relates to a biomedicine event trigger word identification method based on characteristic automatic learning. The biomedicine event trigger word identification method comprises the following steps of 1, data pre-processing; 2, construction of an event trigger word dictionary; 3, construction of candidate trigger word examples; 4, characteristic learning by means of a convolutional neural network model; 5, training by means of a neural network model; and 6, classification of event trigger words. The biomedicine event trigger word identification method is advantaged in that 1, complex preprocessing to data is simplified, and tedious steps for carrying out a characteristic design by people are saved; 2, domain knowledge is introduced, and a lot of external resources such as unlabeled linguistic data are effectively utilized; 3, characteristic automatic learning is carried out by means of a convolutional neural network, manual intervention is reduced, sentence level characteristics in a deeper level can be excavated and explored, through the fusion of local characteristics, implicit global characteristics are discovered, and the category of trigger words can be identified; and 4, a better experiment result is obtained in MLEE linguistic data, and the whole performance on event trigger word detection is improved.

The invention relates to an event extraction system and method oriented to an open domain. The system comprises a preprocessing module, a trigger word identification module, an event parameter identification module, an event atlas analysis module and an event extraction display module, wherein the preprocessing module preprocesses original data information; the trigger word identification module carries out trigger word identification on the basis of a convolutional neural network; the event parameter identification module carries out event parameter identification on the basis of a graph model, the extraction work of an event parameter is converted into a specific graph segmentation problem, and the event parameter is obtained through segmentation; the event atlas analysis module analyzes trigger word identification results and event parameter identification results to obtain the same kind of events; and the event extraction display module carries out visual display on an analysis result so as to bring convenience for users to obtain information. By use of the system, the difficulty that news information can be quickly obtained under a big-data environment is solved, and the user can obtain a news event related to a keyword according to the keyword input on the own so as to provide great convenience for information acquisition.

The invention relates to an identification method, in particular to a biomedical event trigger word identification method based on a syntactic word vector. The biomedical event trigger word identification method comprises the following steps of: 1, pre-processing un-marked data; 2, carrying out word vector training based on syntactic context information; 3, constructing a candidate trigger word dictionary; 4, constructing a trigger word semantic feature vector; 5, training a deep learning model; and 6, identifying a biomedical event trigger word. According to the biomedical event trigger word identification method, syntactic information of the trigger word is precisely acquired by utilizing a larger number of trained word vectors capable of obtaining unmarked data, and input characteristic dimension is effectively reduced; concealed features among the input features are leaned by utilizing the deep learning model, so that the input features are sorted more precisely; and finally, fine adjustment is carried out on word vector information in a training process, so that the word vector information is more suitable for a data set, and thus, the generalization ability and the trigger word identification word of the model are effectively improved.

Identification of inheritance-by-descent haplotype matches between individuals is described. A set of tables including word match, haplotypes and segment match tables are populated. DNA samples are received and stored. A word identification module extracts haplotype values from each sample. The word match table is indexed according to the unique combination of position and haplotype. Each column represents a different sample, and each cell indicates whether that sample includes that haplotype at that position. The haplotypes table includes the raw haplotype data for each sample. The segment match table is indexed by sample identifier, and columns represent other samples. Each cell is populated to indicate for each identified sample pair which position range(s) include matching haplotypes for both samples. The tables are persistently stored in databases of the matching system. As new sample data is received, each table is updated to include the newly received samples, and additional matching takes place.

The invention discloses a Chinese word-dividing method and system in the Chinese medicine information processing domain, which comprises the following steps: A. doing atom cutting for the input Chinese test; building initial cutting word pattern of atom sequence; B. cutting the dictionary and specific word identification based on atom sequence; adding respectively individual word-dividing result into the cutting word pattern; C. generating an optimum word-cutting path according to the word-cutting result in the cutting word pattern; outputting the synthetic word-cutting result according to the optimum word-dividing path. The invention improves the accuracy of Chinese word-dividing with high efficiency, which can identify each kind of specific word selectively according to specific condition.

The invention relates to an integrated method for filtering a short message by introducing query software, which realizes an integrated multi-filtration mechanism by introducing a concept of identifying the short message based on the query software and combining the advantages of conventional technological means. In the method, a sensitive word identification range can be efficiently enlarged by replacing the keyword query part with pinyin; the pertinence is higher by setting a black list and a white list; and the introduced query software can further identify the property of the short message by analyzing the source of the short message. The novel and integrated method improves the accuracy of identifying spam short messages, reduces the error deletion rate, and has the advantages of simple and easy operation and capacity of enlarging the identification range by querying the source of the short message and dividing numbers according to regional groups.

The invention provides a method for automatically correcting an identification error of repeated words in Chinese pronunciation identification. The method comprises the following steps of: (1) performing similarity matching on word confusion networks which are obtained after identification of each sentence, word groups in a word group library and intermediate identification results, and searching the repeated word groups, wherein each word confusion network is a set of all possible identification results and comprises an optimum identification result, namely the original optimum identification result, and the intermediate identification result which corresponds to each word in the optimum identification result, and the word group library comprises the word groups and the intermediate identification results which correspond to the word groups; (2) according to word group information which is obtained by searching, re-calculating a similar probability value and a word identification probability value; (3) according to a new probability value, sorting the word confusion networks according to the size of the probability value; and (4) replacing the optimum identification results and the intermediate identification results of the word confusion networks by using a sorting result. The method has the advantages that: by using experience knowledge in the corrected identification result, the identification error of the repeated words in the current identification sentence is automatically corrected, so the correction efficiency and correction speed of the identification error are improved.

A method is presented to address quantitative assessment of word identification latency of a subject, where the method comprises the steps of: (1) presenting at least one scene, comprising a plurality of letters and a background, to a subject on a display; (2) moving the plurality of letters relative to the scene; (3) receiving feedback from the subject via the input device; (4) quantitatively refining the received feedback; (5) modulating the movement of plurality of letters relative to accuracy of the quantitatively refined feedback; (6) calculating a critical threshold parameter; and (7) recording a critical threshold parameter onto a tangible computer readable medium. An apparatus for quantitative assessment of word identification latency of a subject comprising a display device, an input device, a control device, and a tangible computer readable medium. In its simplest sense, a quantitative assessment profile of word identification latency by psychophysical responses is generated on a tangible computer readable medium.

The invention discloses a word identification method, a word identification device and a word identification system on the basis of the multi-word continuous input. The method comprises the following steps of: detecting a continuous input track on a keyboard region and acquiring track feature data of the continuous input track, wherein the continuous input track is related to a position of a sequence of code characters to be input by a user on the keyboard layout; and matching out at least one candidate item matched with the track feature data of different track sections in the continuous input track from a preset word bank. According to the method, the input efficiency can be improved and the complexity of the user operation is reduced.

An intelligent pair of glasses includes a frame, a pair of glasses held by the frame, an input unit, a camera unit, a projection device rotatably coupled to the frame, and a processor. The input unit generates a control signal for taking photos in response to user inputs. The processor includes a camera control module, a word identification module, a translation module, and a display control unit. The camera control module activates the camera unit and controls the camera unit to capture images according to the control signal to capture an image showing words in an initial language. The word identification module identifies the words shown in the in the image. The translation module translates the words identified in the initial language into words in a target language. The display control unit controls the projection device to project the translated words in the target language onto a surface.

The invention relates to a method and a terminal for rectifying deviation of a file. The terminal comprises an image acquisition and processing module, an image inclination angle detection module and an image deviation rectifying and correcting module, wherein the image acquisition and processing module takes a picture of a file through a camera, acquires an image and processes the file image to obtain a black and white two-dimensional image; the image inclination angle detection module detects the inclination angle of the file image by an image inclination angle detection algorithm; and the image deviation rectifying and correcting module performs deviation rectifying on the image of a visiting card to obtain a corrected image of the visiting card according to the inclination angle of the image of the visiting card detected by the image inclination angle detection module. According to the technical scheme, the deviation of the file image is finally rectified through OCR word identification, text category identification and information checking, amending and authentication, so the method and the terminal overcome the defect of low scan information identification rate existing in the prior art and improve the identification accuracy of the information.

The invention discloses a method and device for conducting voice keyword search. According to the method, at least two kinds of language models are configured in a model file, and each language model comprises an identification module and a corresponding decoding model. The method comprises the step of receiving voice data to be processed, conducing voice feature extraction on the voice data to be processed, using identification models in the model file one by one for conducting language matching on the extracted voice features, determining the identification model with the highest matching rate, determining the decoding model corresponding to the identification model with the highest matching rate from the language models, using the determined decoding model to decode the extracted voice features, obtaining a decoded word identification result, matching keywords in a keyword dictionary and the word identification result, and outputting successfully-matched keywords. According to the scheme, keyword search of at least two languages can be supported, and cost is saved.

The invention discloses a prosodic hierarchy labeling method. The method comprises the steps as follows: acquiring to-be-labeled text data and audio data which have a corresponding relation; extracting a to-be-labeled text characteristic set of each word according to the to-be-labeled text data; extracting an acoustic characteristic set of each word according to the audio data; acquiring a prosodic hierarchy structure by a prosodic hierarchy labeling model according to a word identification of each word, the to-be-labeled text characteristic set of each word and the acoustic characteristic setof each word. The invention also discloses a model training method, a prosodic hierarchy labeling device and a model training device. The prosodic hierarchy labeling model is established by combination of text characteristics and acoustic characteristics, richer characteristics can be provided for prosodic hierarchy labeling, the prosodic hierarchy labeling accuracy can be improved, and the voicesynthesis effect can be enhanced.

The transmitting apparatus includes an encoder creating an encoded content signal by encoding the content, a generator generating sign language word identification information corresponding to chronologically-ordered sign language words appearing in a speech in the content, a creating unit creating control information containing the generated chronologically-ordered sign language word identification information, a storage unit storing sign language word images for displaying a sign language video corresponding to the sign language words by grouping the sign language word images into a plurality of modules according to a frequency of appearance of the sign language words in the speech in the content, a multiplexer creating a data stream by combining the encoded content signal with the control information and by repeatedly replicating the plurality of modules at a frequency corresponding to the frequency of appearance, and a transmitter transmitting the created data stream.

The present invention discloses a system and a method for high concurrent ticket identification based on deep learning. According to the invention, a unified API interface and a ticket classification system are combined so that the system has high compatibility with any ticket input. The combination of an Nginx load balancing server, an HTTP SERVER cluster, a queue server and a GPU ticket identification cluster makes the ticket recognition system high concurrent. The combination of a template adaptation and sequence location system and word identification system of deep learning makes the ticket recognition system easy to operate. The combination of the ticket classification system, the template adaptation and sequence location system, the word identification system of deep learning, the ticket field matching semantic analysis system, the ticket subclass extraction semantic analysis system and the service field content correction semantic analysis system makes the ticket identification system have a high identification rate. Compared with the traditional ticket identification system, the system and the method of the invention have the advantages of good compatibility, high concurrency, easy operability and high identification rate.

The invention discloses a Chinese text keyword extraction method based on document theme structures and semantics, and relates to keyword extraction. The method includes the steps: text preprocessing;Chinese segmentation and part-of-speech tagging; stop word filtering and part-of-speech filtering; keyword extraction. The basic conception of text keyword extraction, Chinese segmentation and English segmentation differences and a common Chinese text keyword extraction method are introduced. A method based on the document theme structures and a method based on semantics are researched, and the principle and an existing implementation scheme are analyzed. In order to overcome difficulty in new word identification in Chinese segmentation, Chinese segmentation effects are continuously improvedby the aid of a dynamically updated segmentation dictionary. The method based on the document theme structures is improved, and global keywords are extracted. Semantic similarities of Chinese words are taken into account, and an algorithm is further improved. The improved algorithm is verified in a self-built data set, good results are acquired by verification experiments and comparison experiments, and keyword extraction effects can be improved by the improved algorithm.

The application provides a voice enhancing method, device, intelligent voice equipment and computer equipment; the method comprises the following steps: obtaining a to-be-processed voice signal; inputting the voice signal into a voice enhancing model, removing noises and / or interference voices in the voice signal so as to obtain a processed voice signal; the voice enhancing model is obtained by training mixed voice signals; the mixed voice signals refer to signals obtained by adding noises and / or interference voices in pure wakeup word voice signals; the pure wakeup word voice signals refer towakeup word voice signals with a noise and / or interference voice ratio smaller than a proportion threshold; the voice enhancing model can effectively remove noises and / or interference voices from thevoice signal, such as voices related to non-wakeup words, thus improving the voice enhancing effect; the method and device can carry out voice identification treatment for the processed voice signal,thus improving wakeup word identification accuracy and wake up efficiency, and improving the user usage experiences.

Techniques for action prediction based on interactive history and context between a sender and a recipient are described herein. In one embodiment, a process includes, but is not limited to, in response to a message to be received by a recipient from a sender over a network, determining one or more previous transactions associated with the sender and the recipient, the one or more previous transactions are recorded during course of operations performed within an entity associated with the recipient, and generating a list of one or more action candidates based on the determined one or more previous transactions. The invention is characterized in that the one or more action candidates are optional actions recommended to the recipient, in addition to one or more actions required to be taken in response to the message. Key word identification out of voice applications as well as guided actions has also been applied to generate action prediction candidates interactive history links. Other methods and devices are also described.

The invention provides a voice identifying method based on deep neural network characteristic training. The method involves the realization and identification of Gabor filter bank characteristics, Gabor filter sub-banks, and a deep neural network (DNN), which is achieved through the following steps: a Gabor filter extracting automatic voice identifying characteristics from a voice signal, firstly on the basis of a distributive voice identification standard, extracting a logarithm Me 1 spectrogram from the voice signal, then conducting convolution on the spectrogram and each 2D filter from the Gabor filter bank; selecting a specific modulation frequency, such that a transfer function of the filter exhibits constant overlapping in a modulated frequency field; an automatic voice identifying system, on the basis of character error rate in a test set, carrying on evaluation, and finally acquiring an identification result. According to the invention, the Gabor filter sub-bank can reduce character and word identification errors, and exhibit the channel distortion resistance and low signal-to-noise ratio. The method uses a voice identifier having a high time modulation filter, has low error rate, and increases the distinctiveness among object types.

The invention discloses an isolation word identification method based on a double-layer GMM structure and VTS feature compensation. The method comprises a training stage and an identifying stage. In the training stage, by voice feature extracting under a pure environment, two GMM training models and an HMM training models are obtained. Each GMM model comprises a GMM1 model containing a small number of Gauss mixing units and a GMM2 model containing a large number of Gauss mixing units. During a noise estimation process at a vector Taylor series (VTS) feature compensation stage, the GMM1 model is used for obtaining the mean value and the variance of noise, a GMM2 model is used for obtaining a pure feature parameters by mapping, and matching with the HMM module is carried out to obtain the final identification results. Compared with an isolation word identification algorithm based on a single GMM model and VTS feature compensation, under the situation that the error recognition rate is not changed basically, noise mean value and variance estimating time is shortened by 90%, feature compensation overall time is shortened by 30%-50%, and calculated quantity of the isolation word identification algorithm based on the VTS feature compensation is effectively lowered.

The invention relates to a synonym identification method, which comprises the following steps: according to a description text to be tested, using an attribute word identification model to obtain the attribute words of the description text to be tested and types corresponding to the attribute words; according to the attribute words and the types corresponding to the attribute words, combining with a user behavior log to calculate relevance characteristics among attribute words; according to the relevance characteristics among sample attribute words selected from the attribute words and textual characteristics among the sample attribute words, training a synonym identification model to obtain the synonym identification model; and according to the relevance characteristics among the attribute words to be tested and textual characteristics among the attribute words to be tested, using the synonym identification model to identify whether all attribute words to be tested are synonyms so as to carry out subsequent processing. According to the technical scheme of the synonym identification method, the comprehensiveness and the accuracy of synonym identification can be improved so as to improve the accuracy and the efficiency of a retrieval result.

The invention provides a data searching method and a device thereof. The data searching method comprises the following steps: obtaining the search key word input by a user, and searching a word index module according so as to key words to obtain the preset word identification information corresponding to the search key word in the word index module; searching an inverted index module according tothe word identification information so as to obtain the preset compressed and stored document information corresponding to the work identification information in the inverted index module; decompressing the document information so as to obtain the document identification information of the document corresponding to the search key word; searing a work position index module according to the search key word and the document identification information so as to obtain the position information of the search key word in the document corresponding to the document identification information; and displaying the document according to the document identification information and the position information. The data searching method and the device thereof can reduce the hardware resource consumption of the searching system and improve the searching efficiency.

The embodiment of the invention discloses a field word identification method and device. In the scheme provided by the embodiment of the invention, a search engine serves as the basis, and the field key word of a field to which a field word to be identified possibility belongs is determined according to the search result of the field word to be identified by the search engine; the score of the field word to be identified, which belongs to the field, is calculated according to the information of the pre-determined field key words and the search result; the score is compared with the field conformity threshold value of the field; and according to a comparison result, whether the field word to be identified belongs to the field is determined. The scheme provided by the embodiment obtains linguistic data having great correlation degree with the field word to be identified by using the characteristics of the search engine, thereby greatly improving the identification speed and accuracy of the field word.

The invention provides a colloquial word identification and semantic identification method and device. The colloquial word identification method includes the steps: acquiring first trained language model; extracting contextual features of words in statements to be identified; identifying the contextual features of words by the first trained language model; determining whether the words are colloquial words or not. The first language model learns the contextual features of the colloquial words in the statements in advance, the identification accuracy of colloquial words in ask questions are improved, the first language model is completed by training, the contextual features of the words in statements to be identified are extracted and identified, so that the method can determine whether thewords are the colloquial words or not, the identification efficiency and the identification accuracy of the colloquial words are improved, and the method solves the technical problems of more colloquial words in the ask questions of the user and low identification efficiency and identification accuracy in the prior art.

The invention provides an intelligent translation earphone system which comprises a processor, a storage module, a communication module, a matching module, a display module, a microphone, a loudspeaker and a power switch, wherein a power supply is connected with the process through the power switch to supply power; the communication module and a network server perform information interaction to realize language recognition, word identification and word translation; the microphone sends an acquired voice to the network server through the communication module; and the storage module, the matching module and the display module all perform information interaction with the processor. The intelligent translation earphone system has the beneficial effects of being reasonable in design and being capable of realizing translation in time and guaranteeing smooth communication.

The invention provides a test paper handwritten English character recognition method and device based on deep learning, and belongs to the technical field of image recognition. The method comprises the steps of acquiring a to-be-recognized test paper image; cutting the acquired image to obtain a word image in the test paper image, and identifying the word image by using a trained neural network model based on an attention mechanism to obtain a word identification result; wherein the step of cutting the acquired image specifically comprises the sub-steps of carrying out binarization operation on the test paper image, cutting text lines in the test paper image, and cutting English words in the text line image; according to the method, the English text line segmentation method based on dynamic line segmentation and the word sequence recognition method based on the attention mechanism are adopted, a good segmentation effect is achieved on the bent text, and the word recognition accuracy iseffectively improved.

The invention provides a method and a device for new word identification. The method comprises that a plurality of unmatched continuous segments are extracted from a search query word submitted by a user; statistics of corresponding relations between contents of a clicked search result web page corresponding to the search query word and the multiple segments is carried out; and according to the corresponding relations, whether the continuous multiple segments in the search query word will be identified into a new word is judged. According to the method and the device for the new word identification provided by the invention, whether the segments of the search query word can form the new word can be analyzed according to the statistics and analyze of the corresponding relations between the segments in the search query word and the search result webpage.