53 results about "Return statement" patented technology

A user statement associated with a query is received. The user statement is parsed to determine a set of interpretations matching the user statement. Based at least in part on the set of interpretations the query is determined to be a candidate for iterative improvement. The query is iteratively improved at least in part by prompting a user associated with the user statement to provide a further input.

The invention discloses a method for image automatic text annotation based on generative adversarial networks, which comprises the steps of generating false statements by a generator, rebuilding a discriminator at the same time, and inputting the generated statements and real statements into the discriminator to train until the discriminator cannot discriminate the real statements and the generated statements. The method disclosed by the invention changes a problem that sentences are rigid and inflexible in CNN-RNN image automatic statement annotation, and enables the generated sentences to be more accurate, more natural and more diversified. The generated statements can face more complex scenes in reality and more conform to language expression means annotation images of the human, thereby having more extensive applications in practice.

Embodiments of the invention disclose translation model-based training and translation methods and apparatuses, and a storage medium. The training method comprises the steps of inputting a source statement to a translation model, thereby obtaining a target statement output by the translation model; determining the fidelity of the target statement to the source statement; by taking the target statement and a reference statement as inputs of a judger model and taking the fidelity as an output of the judger model, training the performance of calculating the similarity between the target statementand the reference statement by the judger model; outputting the similarity between the target statement and the reference statement by using the judger model; and by taking the source statement as aninput of the translation model, taking the target statement as an output of the translation model and taking the similarity as a weight coefficient, training the performance of outputting the corresponding target statement according to the input source statement by the translation model.

The invention discloses a method for realizing non-barrier automatic psychological consult by adopting a computer, relating to a psychological consult technology and solving the problems of unavailable independent operation, consult and monitoring without real-time feedback, consult results without duly feedback, psychological consult files without establishment of overall process of consult and therapy, and weak expansibility of traditional psychological consultant adjuvant therapy equipment. A process of the method comprises the steps of: A, carrying out extraction of key words, expansion key words and work string sequence on input natural language statements; B, carrying out semantic matching, wherein the semantic matching comprises key words matching, expansion key words matching and word string sequence matching, and if certain matching in the semantic matching exceeds a threshold value, the matching is judged to be successful, or else, information incapable of being answered is output; and C, carrying out conversation management and outputting return statement, wherein the conversation management comprises rehearsal, a content reflection technology and conversation context management. The invention can be used independently or mutually compensated with a psychological consultant, which makes up the deficiency of the psychological consultative organization to some extent.

The invention provides a statement generation method, which comprises the following steps that: obtaining a target statement element, wherein the target statement element is generated in a way that the statement element used for displaying real-time data is customized in a statement generation template, and the statement generation template can customize a development statement; on the basis of the target statement element, obtaining real-time data to be displayed; and injecting the obtained real-time data to be displayed into the target statement element so as to generate a target statement used for displaying the real-time data to be displayed. The invention also provides a statement generation system, a computer readable medium and a computer system.

The invention discloses an answer extraction method and system based on deep learning. wherein an implementation of the method includes the steps of collecting knowledge fragments related to the answer of a question to be answered according to keywords in the question to be answered; based on the type of the question to be answered, analyzing the part-of-speech tagging of the knowledge fragment, and using the knowledge fragment containing the type of the question to be answered as a candidate answer; using the trained word2vec model to calculate the similarity between the keywords and words contained in the knowledge fragment in the candidate answer, and finding words that meet the similarity requirements as target candidate answers; substituting the words in the target candidate answers into the question to be answered to obtain a number of target statements, sorting each target statement by judging semantic information of each target statement, and using the target statement with thehighest score as the target answer. Through the extraction method and system, the answers of multiple words with higher similarity can be found, thereby improving the accuracy of the system.

The invention provides a statement analysis method and apparatus, a computer device and a readable medium. The method comprises the steps of obtaining feature information of heuristic understanding corresponding to a to-be-analyzed target statement; according to the feature information of the heuristic understanding corresponding to the to-be-analyzed target statement, obtaining multiple candidateanalysis results of the target statement and confidence degrees corresponding to various candidate analysis results, wherein each candidate analysis result comprises a candidate intention and a candidate slot position; and according to the candidate analysis results and the corresponding confidence degrees, obtaining target understanding corresponding to the target statement. According to the statement analysis scheme, even in a cold-start stage with insufficient training data resources, the feature information of the heuristic understanding corresponding to the target statement can be obtained, thereby realizing accurate understanding of the target statement. Moreover, a template in a fixed format can be prevented from being used for performing statement identification, so that the technical defect of poor statement understanding accuracy caused by weak template generalization capability can be avoided and the statement analysis accuracy and efficiency can be effectively improved.

An embodiment of the invention discloses an annotation library generation method and an annotation library generation device. The method includes: aligning a target statement and a source statement, mutually translated in sentence level, in a bilingual parallel corpus; if a first specific word, belonging to a first word class, in the target statement has no corresponding source specific word in the source statement, acquiring a first candidate set which includes candidate source specific words in mutual translation relation with the first specific word; acquiring a candidate position set according to the alignment relation between the source statement and the target statement, wherein the candidate position set includes positions of possibly lost source specific words in the source statement; acquiring the correct probability of each statement in a second candidate set according to a preset language probability model, wherein the second candidate set includes candidate source statements formed by filling of the candidate source specific words in the first candidate set to the positions of the candidate position set; generating an annotation library which includes candidate source statements determined according to the correct probability of statement in the second candidate set.

The invention discloses a Chinese text proofreading method based on a knowledge graph. Firstly, an entity extraction technology is used for extracting to obtain an entity in a text statement, and thenaccording to a matching result of the entity and a relation rule, a syntactic semantic error type is searched and discovered in a knowledge graph. According to the method, dependence on a large-scaleword library can be avoided, and semantic proofreading is conducted on texts from the three aspects of wrongly written characters, component deletion and definition contradiction. Compared with an existing Chinese automatic proofreading system, the proofreading method has high recall ratio for proofreading various semantic errors in the limited field. The method can effectively improve the accuracy and recall ratio of text proofreading, and is helpful for text workers to improve the text quality.

The invention discloses a translation method and system, which is used for improving translation efficiency. The translation method provided by an embodiment of the invention comprises the steps of matching a to-be-translated statement with a source statement in a preset cache; if the matching succeeds, outputting a target statement corresponding to the successfully matched source statement, wherein a corresponding relationship between the source statement obtained in a previous translation task process and the target statement is pre-stored in the cache, and the source statement and the target statement are translation results for each other.

Scoring and detecting anomalies in a dynamic number of transaction records within a relational database management system by initiating a function for accepting and storing a transaction record in the relational database management system; This function and the analytics are integral to the RDBMS, as distinguished from extracting the data and passing the data to a separate application external to the RDBMS. The function determines if the transaction is for a different individual then a previous transaction or for the same individual. If the transaction is not for a different individual, adding the new transaction data to a memory work space and returning a NULL statement, otherwise passing the memory work space to a service. In the next step the memory workspace is accepted and an analytic model is executed to produce a score, and return the score to a calling function. This score is received at the calling function, the memory work space for the individual is released, and the score is returned to the calling statement. All of the NULL statements are collapsed and a single record is retained for the individual. This record contains the score, that is, a buying preference or pattern, a travel pattern, a calling pattern, a maintenance prediction, a risk score, or a credit score, by way of illustration.

The embodiment of the invention discloses a statement generation method and device, and improves statement development efficiency. The method comprises the following steps: obtaining the identifier of a target statement; according to the identifier of the target statement, looking up a statement parameter corresponding to the identifier of the target statement from a pre-stored statement parameter mapping table; utilizing the statement parameter to generate a SQL (Structured Query Language) statement; and executing the SQL statement to generate the target statement.

A passenger management apparatus can appropriately manage the return state and number of passengers and prevent getting-on of a person not scheduled to get on such as a suspicious person.

The invention provides a new energy policy information extraction method and system. The method is characterized by processing a new energy policy into a new energy policy text in a set format; basedon a pre-established new energy policy feature dictionary library, using a punctuation symbol automatic identification sentence division method to perform sentence division and marking on the new energy policy text, outputting a division and marking result, and generating a text sentence library; and based on a unified service interface, generating key information of a new energy policy feature dictionary library and a text statement library according to the text statement and outputting the key information.

This invention provides a analysis method and system for protocol information composed of some IEs. Wherein, the method comprises following steps: 1) dividing and selecting a set rule signs for control action from all logic control actions used in protocol analysis process, the control action comprises: positioning IE position, connecting and uniting IE, calculating, deciding and outputting the IE; 2) generating a general explainer to recognize the said rule signs; 3) setting rule database unit according to a exact protocol; 4) when receiving the bit stream, the said explainer reads rule statement from rule database, executes every control action set by rule sign, and completes the analysis. When the protocol changes, it needs just to modify the rule database unit instead of the said explainer; thereby, the invention decreases modification time and analysis difficulty.

The invention discloses a full-automatic closed-loop control method of a wireless network work order. Source data are collected from an underlying data source through Hadoop and are classified and calculated through a preset classification algorithm, index values of the source data are obtained, then the index values are compared with a preset threshold value to obtain problem data, the problem data meeting the order sending condition generate a work order, the work order is stored in a work order table, a big data system and an electronic operation and maintenance system carry out order sending, order statement and filing on the work order in the work order table respectively, and therefore the closed-loop control over the full work order sending, shifting and settlement process is achieved. Thus, the operating quality and efficiency of the wireless network are improved, and meanwhile work accuracy and objectivity are ensured.

An illegal pattern and a computer program having a method are received. The method has one or more return statements, and a number of basic blocks. The method is normalized so that each return statement of the target method relating to the illegal pattern returns a constant Boolean value. A first path condition and a second path condition for one or more corresponding paths is determined such that one or more corresponding basic blocks return a constant Boolean value of true for the first path condition and a constant Boolean value of false for the second path condition. An unsatisfiability of each path condition is determined using a monadic second-order logic (M2L) technique. Where the unsatisfiability of either path condition is false, the method is reported as not being a validator. Where the unsatisfiability of either path condition is true, the method is reported as being a validator.

The invention discloses a composition processing method and system based on syntactic analysis, and the method comprises the steps: carrying out the text data preprocessing of a composition text to becorrected, and obtaining the preprocessed target text data; judging whether an error exists in a logic relation between a target statement and an adjacent statement in the target text data or not, ifyes, identifying a first abnormal statement corresponding to the target statement and the adjacent statement in the composition text to be corrected, and performing first marking on the first abnormal statement; and/or comparing the syntactic structure of the target statement in the target text data with a standard syntactic structure of a preset standard statement in a standard regular databaseto judge whether the syntactic structure of the target statement conforms to the standard syntactic structure or not; if not, identifying a second abnormal statement corresponding to the target statement in the composition text to be corrected; and the second abnormal statement is marked and corrected, so that the grammatical error in the composition can be automatically corrected, and the composition correction efficiency is improved.

The invention provides a method and device for classifying services based on at least one service classification model. The service classification method comprises the steps aiming at each service classification model in the at least one service classification model, providing representative user statement samples corresponding to to-be-classified user statements and the service classification model as statement pairs to the service classification model to determine the similarity between the to-be-classified user statements and the representative user statement samples so as to determine thesimilarity between the to-be-classified user statements and all service categories; and determining the service category of the to-be-classified user statement based on the similarity between the to-be-classified user statement and each service category, wherein each service classification model in the at least one service classification model corresponds to one service category. According to themethod and the device, the service category of the to-be-classified user statement is determined based on the at least one service classification model for each service category, so that the service classification failure caused by ignoring some service categories can be avoided.

The invention discloses a synchronous termination method and system of wide area fault recording. Fault recording devices are arranged in each substation in the system and are grouped according to a power supply system structure, and the devices in the same group are accessed to a satellite timing system and are connected via a communication network for communication. The method comprises the following steps: after the fault recording devices are started or return, sending starting signals or return signals to the other devices in the same group, and when the fault recording devices return orreceive network return signals, starting to wait for a time length delta T; marking the effective positions 1 according to the return state of the fault recording devices in the group during the waiting; if the effective positions are all 1 after the waiting, terminating the recording at the latest return moment in the return signals in the group; and if not all effective positions are 1, terminating the recording at the latest return moment of the delta T in the group. According to the synchronous termination method and system, synchronous communication can be achieved by using the constructed non-real-time E1 network, the synchronous termination of wide area fault recording can be achieved, and more complete data support is provided for the system fault analysis.

The application discloses a word alignment training method, a machine translation method and system for increasing the efficiency of word alignment training. The application provides a word alignment training method comprising following steps: determining a vocabulary translation general table of multiple parallel corpus, wherein the vocabulary translation general table comprises the translation probability of a word of a source statement to a word of a target statement in the parallel corpus; splitting the vocabulary translation general table to obtain multiple vocabulary translation sub-tables, wherein the vocabulary translation sub-table comprises the translation probability of at least one word of a source statement to a word of a target statement in the parallel corpus; on the basis of the vocabulary translation sub-tables, determining the alignment relationship between words in the source statement and words in the target statement in the parallel corpus.

The invention discloses an emulator with the program counter value settable. The emulator comprises an emulation chip, a debugging module and a storage. A user program and data storage area and a monitoring program storage area are arranged inside the storage. The debugging module controls the emulation chip to be in a monitoring mode or in a user mode, and in the monitoring mode, the emulation chip is controlled to execute a program at the position of a corresponding address in the monitoring program storage area, wherein a PC value correcting monitoring program is stored in the monitoring program storage area. When a return statement of the PC value correcting monitoring program is executed, the current PC value automatically becomes an address value stored in a stack. Under the situation that a processor core of the emulation chip does not support PC value instruction writing, the emulator can still provide the debugging function of correcting PC values; the user program debugging efficiency is improved, meanwhile, the dependence of emulation debugging function realization on the functions of the processor core is lowered, the development complexity and the cost of the emulation chip are lowered, and the universality of emulator design is improved.

The invention discloses a text classification method based on a word sense disambiguation convolutional neural network. The method comprises the following steps of configuring an ambiguous word bank with determined word senses; obtaining a related file, extracting text content from the file, and performing word segmentation processing on each statement in the text; determining the part-of-speech of each word in the statement; determining a disambiguation target word; determining the meaning of the target word and performing disambiguation processing; performing word segmentation processing andstop word removal processing on original statements contained in the disambiguated text to obtain target statements corresponding to the original statements; determining the criticality of words in the target statement; determining the criticality of the target statement; sorting the statements according to the criticality of the statements to obtain a target text; and utilizing the trained textclassification model based on the convolutional neural network to classify the target text. According to the method, text classification can be carried out based on the word sense disambiguation convolutional neural network, the text classification method is optimized, the text classification efficiency and the text classification accuracy are improved, and time and labor are saved.

The embodiment of the invention provides a target statement identification method and device. The method includes the steps of obtaining to-be-processed text which includes one or a plurality of natural language statements; extracting recognition features of each statement, wherein the recognition features include a first feature and/or a second feature, the first feature is used for indicating features of the statement in the semantic aspect, and the second feature is used for indicating features of the statement in the text aspect; and identifying target statements in the text according to aprebuilt target statement recognition model and the recognition features of each statement in the text. According to the target statement identification method and device, statements belongs to the target statements (such as graceful statements) can be automatically found, and thus target statement identification efficiency is greatly improved; meanwhile, the identification standard is based on objective features and models, so that identification results are objective, and thus the subjectivity problem during manual identification is solved.

The invention provides a database operation method and device. The method comprises the following steps: receiving a service processing request; obtaining an XML configuration file corresponding to the service processing request, wherein a processing logic statement related to the service processing request is recorded in the configuration file, the processing logic statement at least comprises two types of a database operation statement and an operation result return statement; and reading and executing the database operation statement in the XML configuration file to obtain a database operation result, and reading and executing an operation result return statement in the XML configuration file to return the obtained database operation result. Compared with an existing mode of performingdatabase operation by using a JAVA file, the JAVA file is not used, the database operation statement is recorded in the XML configuration file, and the execution efficiency of the XML configuration file is superior to that of the JAVA file, so that the execution efficiency of the method and the device is higher than that of an existing database operation mode.