65 results about "Coreference" patented technology

Coreference resolution is the process of identifying when two noun phrases (NP) refer to the same entity. Two main contributions to computational coreference resolution are made. First, this work contributes a new method for recognizing when an NP is anaphoric. Second, traditional approaches to coreference resolution typically select the most appropriate antecedent by recognizing word similarity, proximity, and agreement in number, gender, and semantic class. This work contributes a new source of evidence that focuses on the roles that an anaphor and antecedent play in particular events or relationships. I show that using contextual role knowledge as part of the coreference resolution process increases the number of anaphors that can be resolved, and I demonstrate an unsupervised method for acquiring contextual role knowledge that does not require an annotated training corpus. A probabilistic model based on the Dempster-Shafer model of evidence is used to incorporate contextual role knowledge with traditional evidence sources.

A method and a system for coreference resolution are provided. The method includes receiving a set of document clusters, each cluster in the set of document clusters including a set of text documents. Instances of each of a set of candidate named entities are identified in the document clusters. For a pairs of the candidate named entities, at least one socio-temporal feature is computed that is based on the similarity of the distributions of identified instances of the respective candidate name entities among the document clusters. A decision for merging for the candidate named entities into a common real named entity is based on the socio-temporal features.

The invention discloses a method and a system for fusing event information. The method and the system are used for extracting, replenishing, clustering and fusing the event information to form a complete event with the high integrity degree of the event information. The method comprises the following steps of: generating an original selection event set including a plurality of events; comparing the similarity of the events in the original selection event set and an event extraction mode to form a candidate event set; discriminating and annotating the candidate event set to generate a trainingsample, and generating an inference rule, a zero coreference resolution model, an event identification and extraction model and an argument identification and extraction model of the related events by the training sample; acquiring webpage texts from a webpage of the complete event to be extracted to generate event-annotated texts, and performing structural replenishment on clauses with structural deficiency to generate event-replenished annotating texts; extracting event mentions and event arguments of the event-replenished annotating texts to obtain a first event set; and clustering the event examples of the first event set, and normalizing to generate the complete event.

System, method and program product for text processing using object coreference technology. In particular, the invention provides a text processing method which includes, acquiring text to be processed; extracting subject words and entity words corresponding to the subject words from the text; grouping the subject words; determining entity words that reference a same concerned object according to the grouped subject words; and generating processing policy for entity words that reference a same concerned object. The invention also includes a system with means for carrying out the method. The invention generally realizes automatic, more comprehensive, accurate, efficient analysis and processing on text data. The invention can be used to dig a large amount of comment data about some entity, and the invention can also be used to suggest insertion place in an article where embedded advertisement is inserted.

The present disclosure relates dialog states, which computers use to internally represent what users have in mind in dialog. A dialog state tracker employs various rules that enhance the ability of computers to correctly identify the presence of slot-value pairs, which make up dialog states, in utterances or conversational input of dialog. Some rules provide for identifying synonyms of values of slot-values pairs in utterances. Other rules provide for identifying slot-value pairs based on coreferences between utterances and previous utterances of dialog sessions. Rules are also provided for carrying over slot-value pairs from dialog states of previous utterances to a dialog state of a current utterance. Yet other rules provide for removing slot-value pairs from candidate dialog states, which are later used as dialog states of utterances.

A scalable hierarchical coreference method that employs a homomorphic compression scheme that supports addition and partial subtraction to more efficiently represent the data and the evolving intermediate results of probabilistic inference. The method may encode the features underlying conditional random field models of coreference resolution so that cosine similarities can be efficiently computed. The method may be applied to compressing features and intermediate inference results for conditional random fields. The method may allow compressed representations to be added and subtracted in a way that preserves the cosine similarities.

The invention belongs to the technical field of natural language processing for mathematics, in particular to an extraction, expression and modeling method of text semantics aimed at elementary mathematical questions and a corresponding question meaning analysis system of elementary mathematics. The method includes the following steps: as for an inputted mathematical question, using a combination of a word segmentation lexicon and a regular expression to segment words, as for the result after segmenting the words, conducting word conversion and word group combination, and conducting object replacement of reference words through anaphora resolution; then using the information obtained after processing to extract and translate mathematical formulas by virtue of a first-order logic, obtaining a mathematical question expression based on the first-order logic; finally, using deep neural networks to conduct semantic modeling and semantic fusion to the natural language and formulas of the question. The effective expression and modeling method of elementary mathematical questions proposed by the extraction, expression and modeling method and system of text semantics aimed at elementary mathematical questions can convert the mathematical question to a semantic representation which can be processed by a computer and conduct a more precise semantic modeling of mathematical questions.

A system and method provide for identifying coreference from serialized data coming from different services. The method includes generating a tree structure from serialized data. The serialized data includes responses to queries from the different services. The responses each identify a hierarchical relationship between a respective set of objects. Nodes of the tree structure each have a name corresponding to a respective one of the objects. The tree structure is traversed in a breadth first manner and, for each node in the tree structure, a respective pairwise similarity is computed with each of the other nodes of the tree structure. The computed pairwise similarity is compared with a threshold to identify co-referring nodes that refer to a same entity. The threshold is a function of a depth of the node in the tree structure.

The invention discloses an abstracting method for mass-text quick understanding. The abstracting method includes the steps that the content of text is obtained; the text is subjected to pretreatment operating such as word segmentation, coreference eliminating, redundant information removing and analysis unit dividing; the content of the text is subjected to subject analysis with a topic model to obtain subject distribution in the text; a graph model is built according to subject incidence relationships between analysis units, and weights of all directed edges in the graph model are calculated; the graph model is calculated with the contribution iterative method till is converged, and the text summarization with the suitable length is generated as requirement. According to the text abstracting method, mass unstructured text data can be automatically analyzed, the text summarization with which a core topic can be fully covered is obtained to serve as replacement of mass original data, and therefore the quick understanding aim is achieved.

The invention discloses an instance-based dynamic generalization coreference resolution method, and relates to the field of text information extraction. The dynamic generalization coreference resolution method comprises a training instance library establishment stage and an in-discourse entity resolution stage, and the coreference resolution is finished by instance establishment, instance library establishment, index creation, dynamic generalization, instance retrieval and coreference chain combination. The method eliminates the long tail effect in a coreference statistical model, fully achieves the effect of a low-frequency training sample, makes full use of the precious training sample, and makes the dynamic generalization mechanism of the instances self-adaptively convert the classification of test instances into the selection and utilization of the best generalization point in a training instance library and finally find the optimally matched training instance.

The invention discloses a Chinese zero pronoun resolution method and system. The method includes: obtaining a zero pronoun mark by preprocessing a target language; carrying out position recognition ofcandidate zero pronouns; combining the result of position recognition with the preset optimization rule to get the target zero pronoun.; obtaining a set of expression pairs according to all target zero pronouns and candidate antecedents; obtaining the probability of anaphora relationship between target zero pronoun and candidate antecedents and sorting the probabilities of multiple anaphora relationships; according to the sorting results, obtaining the corresponding zero pronoun resolution results. The invention utilizes preset optimization rules combined with syntactic analysis to realize accurate identification of zero pronouns, and the zero pronouns resolution is completed by using a deep learning method.

The invention discloses a text information processing method and a related device, for improving a pronoun anaphora resolution effect. The text information processing method comprises the steps of determining a first pronoun and a first antecedent in a to-be-processed text; determining a first vector representation value of the to-be-processed text, wherein the first vector representation value isused for representing semantic information of the to-be-processed text; determining a first semantic feature vector corresponding to the first pronoun and the first antecedent; obtaining a first vector representation value and an anaphora prediction result corresponding to the first semantic feature vector through an anaphora prediction model; and if the anaphora prediction result is that an anaphora relationship exists between the first pronoun and the first antecedent, replacing the first pronoun in the to-be-processed text with the first antecedent to obtain a processed text. According tothe text information processing method, on the basis of considering the semantic features between the pronoun and the antecedent, the context semantic information of the pronoun is also fused, so thatthe recognition rate of the anaphora can be effectively improved, and the anaphora resolution effect of the pronoun is improved.

The invention discloses an entity coreference resolution method based on similarity. The implementation process includes the steps that firstly, data in a data set is preprocessed to form data pairs, and the data pairs are entity pairs; secondly, weights are set, and similarity values are calculated and compared with a set threshold; thirdly, when the set threshold is reached, entity unification is carried out, that is, all the data pairs reaching the threshold are fused into one datum; when the set threshold is not reached, data summarization is carried out, and the data pair data is summarized to form a new data set, wherein a summarization result comprises the combined datum and data smaller than the threshold. Compared with the prior art, the weights and measurement indexes of similarity, a good processing effect is achieved, the requirement for entity coreference resolution in massive data processing can be met, effective guarantee is provided for entity coreference resolution, practicality is high, and popularization is easy.

A method and a system for coreference resolution are provided. The method includes receiving a set of document clusters, each cluster in the set of document clusters including a set of text documents. Instances of each of a set of candidate named entities are identified in the document clusters. For a pairs of the candidate named entities, at least one socio-temporal feature is computed that is based on the similarity of the distributions of identified instances of the respective candidate name entities among the document clusters. A decision for merging for the candidate named entities into a common real named entity is based on the socio-temporal features.

The present invention discloses a coreference resolution-oriented multi-semantic web entity contrast table automatic generation method. The method comprises the following steps of: giving a set of candidate coreference entities, and combining attributes with similar semanteme in the set of entities according to structure and textual information first; then, scoring the attributes based on the combined attributes and value distribution of the entities in the attributes, calculating the redundancy of candidate attributes and selected attributes, choosing an attribute with high score and low redundancy to be added into a key attribute set, and repeating the step until a predetermined number of attributes are all selected or no attributes can be chosen; and at last, based on values of key attribute organization entities in key attributes, generating a visual entity contrast table for users to participate in entity coreference resolution. By applying the coreference resolution-oriented multi-semantic web entity contrast table automatic generation method provided by the present invention, the accuracy and efficiency of user participation in multi-semantic web entity coreference resolution are improved.

The invention provides a method for analyzing discourse coherence quality of an English writing. The method refers to an analysis model composed of an English writing preprocessing module, an English writing grammatical role marking module, an English writing feature extraction module, an English writing coreference resolution module, an English writing physical link network construction module and an English writing discourse coherence analysis module. After an English writing is processed by the analysis model, the discourse coherence quality analysis results of the English writing can be finally obtained. According to the method disclosed by the invention, the problem of coherence analysis of reference relation and language words in the English writing is solved, and the analysis result of the method disclosed by the invention is better than that of the traditional method for analyzing the discourse coherence quality of the English writing.

The invention relates to an information processing device and an information processing method. The information processing device comprises a translation relationship obtaining unit, a nominal composition confirming unit, a normalization unit, a structured pattern generation unit and a phase pattern generation unit; the translation relationship obtaining unit obtains translation relationships of corpuses in a bilingual parallel corpora between two languages; the nominal composition confirming unit tags part of speech of the corpuses in the second language and confirms nominal compositions and non-nominal compositions of the corpuses of the two languages; the normalization unit replaces the nominal compositions of the corpuses in the two languages into coreference symbols and accordingly structured corpuses in the two languages are formed; the structured pattern generation unit generates structured patterns between the two languages; the phase pattern generation unit generates phase patterns between the two languages. According to the information processing device and the information processing method, the structured patterns and the phase patterns between the two languages can be provided and accordingly corpus switching between the two languages can be well achieved.

The invention discloses an information processing method and electronic equipment and solves the technical problem that personal pronouns in a voice instruction cannot be correctly identified by the existing electronic equipment. The method is applied to the electronic equipment and comprises the steps that input voice is obtained; the input voice is identified by a voice identification engine; when the condition that the input voice comprises the personal pronouns is identified, first data are obtained; a coreference object which is delegated by the personal pronouns is determined on the basis of the first data; an operation instruction is executed on the basis of the coreference object, wherein the operation instruction is an instruction which corresponds to the input voice and is identified by the voice identification engine after the input voice is identified by the voice identification engine.

Previous neural network models that perform named entity recognition (NER) typically treat the input sentences as a linear sequence of words but ignore rich structural information, such as the coreference relations among non-adjacent words, phrases, or entities. Presented herein are novel approaches to learn coreference-aware word representations for the NER task. In one or more embodiments, a CNN-BiLSTM-CRF neural architecture is modified to include a coreference layer component on top of the BiLSTM layer to incorporate coreferential relations. Also, in one or more embodiments, a coreferenceregularization is added during training to ensure that the coreferential entities share similar representations and consistent predictions within the same coreference cluster. A model embodiment achieved new state-of-the-art performance when tested.

The invention discloses a multi-round dialogue rewriting method and system based on text editing and grammar error correction, and the method comprises the steps: carrying out word-level labeling of a collected dialogue data text through a text labeling algorithm, generating text labeling data, and carrying out fine adjustment of a deformation-based bidirectional coding representation model, so as to obtain a sequence labeling model; editing dialogue history and incomplete statements according to the classification label of each word in the prediction sequence of the model, and finally carrying out grammar error correction modeling on the rewritten text to improve the fluency of the statements. According to the method, the accuracy of multi-round dialogue rewriting can be improved, the problems of anaphora and omission in a dialogue system are effectively solved by utilizing a text editing and grammar error correction method in dialogues, and the integrity of dialogue statements is improved.

The invention relates to the field of artificial intelligence, the invention discloses a method and device for semantic completion in multiple rounds of conversations, equipment and a storage medium.Grammar detection is carried out on the multiple rounds of conversations through a preset corpus sentence segmentation function and a preset analysis function, statements with incomplete semantics arecompleted, the accuracy of semantic analysis results is improved, and the accuracy of searching corresponding response information according to the semantic analysis results is improved. The method comprises the steps that grammar detection is conducted on a first statement and a second statement through a preset corpus sentence segmentation function and a preset analysis function, and a first statement detection result and a second statement detection result are obtained; when the second statement detection result comprises a single entity and the second statement is a questionnaire, supplementing the semantic missing part of the second statement according to the first statement detection result to obtain a first supplemented statement; and if the first complement statement comprises words with unknown indications, replacing the words with unknown indications in the first complement statement according to a first statement detection result to obtain a second complement statement.

A system performs conversations with users using chatbots customized for performing a set of tasks. The system may be a multi-tenant system that allows customization of the chatbots for each tenant. The system processes sentences that may include negation or coreferences. The system determines a confidence score for an input sentence using an intent detection model, for example, a neural network. The system modifies the sentence to generate a modified sentence, for example, by removing a negation or by replacing a pronoun with an entity. The system generates a confidence score for the modified sentence using the intent detection model. The system determines the intent of the sentence based on the confidence scores of the sentence and the modified sentence. The system performs tasks based on the determined intent and performs conversations with users based on the tasks.

The embodiment of the invention provides an anaphora resolution method and device. According to the method, the antecedent candidate set corresponding to each training sample is determined, and the feature vector is constructed for each element in the antecedent candidate set according to the pronoun category in each training sample so as to reflect the semantic relationship between the sentencesand the antecedents, so that the advantages of the semantic relationship can be effectively exerted; and then, the feature vector of each element in the antecedent candidate set and the anaphora resolution result of the corresponding training sample are input into the maximum entropy model for training, so that anaphora resolution can be performed on the statement by adopting the anaphora resolution model obtained by training. Therefore, the context semantic relationship of the anaphora can be fully utilized, so that the semantic relationship between the anaphora and the anaphora can be subsequently and effectively recognized, and the accuracy and recall rate of anaphora resolution are improved.

The invention relates to an event-based Chinese coreference corpus library establishment method. The method mainly comprises the following steps of (1) selecting a CEC2.0 corpus library as a basis of establishment; (2) determining a target and an annotation mode of coreference annotation; (3) making a corresponding annotation specification according to a specific coreference target; (4) performing text preprocessing on CEC2.0 corpora; (5) automatically annotating event elements and event coreference; (6) further optimizing an annotation result through manual annotation; and (7) setting a consistency check step to ensure the quality of corpus annotation. According to the method, the defects of an existing coreference resolution corpus library are overcome; the method not only can cover all events in the corpus library but also is established based on Chinese syntactic analysis and semantic analysis, and conforms to the characteristics of Chinese; and the method also can perform consistence check on annotated corpora to ensure the quality of the corpus annotation.

The invention relates to a Chinese zero-anaphora resolution method and a Chinese zero-anaphora resolution system based on a Mask mechanism and a twin network. The method comprises the following stepsof: adding a [MASK] mark at the position of a zero pronoun; wherein if the antecedent and the [MASK] are in the same sentence, not carrying out splicing processing and if the antecedent and the [MASK]are not in the same sentence, carrying out splicing processing on the sentence where the antecedent is located and the sentence where the complemented zero pronouns are located; inputting the preprocessed sentence into a pre-trained BERT model to extract a first antecedent and a first zero pronoun; integrating an attention mechanism into the BERT model, and processing the first antecedent througha first linear function to obtain a second antecedent; for the first zero pronouns, in combination with preselected manual features, acquiring second zero pronouns through respective linear functionprocessing; and calculating the similarity between the second antecedent and the second zero-generation word, and outputting the antecedent with the highest similarity. According to the invention, information redundancy and noise are avoided.

A system and method provide for identifying coreference from serialized data coming from different services. The method includes generating a tree structure from serialized data. The serialized data includes responses to queries from the different services. The responses each identify a hierarchical relationship between a respective set of objects. Nodes of the tree structure each have a name corresponding to a respective one of the objects. The tree structure is traversed in a breadth first manner and, for each node in the tree structure, a respective pairwise similarity is computed with each of the other nodes of the tree structure. The computed pairwise similarity is compared with a threshold to identify co-referring nodes that refer to a same entity. The threshold is a function of a depth of the node in the tree structure.

The invention provides an anaphora resolution method for a multi-round dialogue system, which comprises the following steps of: S1, detecting statements received by the multi-round dialogue system, judging whether the statements need anaphora resolution or not, and if so, entering the step S2; S2, judging the statements determined to be required to be subjected to anaphora resolution, distinguishing anaphora types of the statements, and screening candidate entities from the statements of which the anaphora types are distinguished; wherein the anaphora type of the statement comprises a back-pointing statement and a co-pointing statement; S3, determining the distances between candidate entities and the demonstrative pronouns in statements, and taking the candidate entities with the smallestdistance as demonstrative linking words; S4, updating the demonstrative pronouns into demonstrative linking words; The demonstrative pronouns of the statements input into the multi-round dialogue system can be accurately recognized, accurate anaphora resolution is carried out, smoothness of the multi-round interaction system can be effectively improved, and user experience is improved.

The invention discloses a chapter-level multi-event extraction method based on argument subgraph prompt generation and guidance. According to the method, a chapter-level long text encoder is used to obtain complete text features, and chapter-level information and sentence-level information can be utilized at the same time. Anaphora and positioning of multiple events are realized by extracting the generated event sketch through the multi-element argument relationship, and the argument classification is realized by performing event slot filling by using a pre-training model method based on a prompt normal form, so that the multi-event extraction accuracy is improved. The method does not need to use a trigger word, and reduces the annotation burden of the data set.

The invention provides an end-to-end multitask learning dialogue anaphora resolution method and a system. The system comprises a context information representation module, a zero pronoun attention representation module, a depth detection model and a replacement module. The context information representation module is used for preprocessing a historical dialogue and a current dialogue, extracting a candidate word context representation and a pronoun context representation, and carrying out attention weight calculation on the candidate word context representation and the pronoun context representation; the zero pronoun attention representation module is used for further carrying out attention weight calculation on the candidate word context representation and the pronoun context representation; the deep detection model is used for judging whether an anaphora phenomenon exists in a current session, and the replacement module is used for replacing pronouns and zero pronouns with candidate words. According to the method, an end-to-end multi-task deep learning technology is adopted, the resolution task is completed based on attention mechanism representation, the resolution accuracy is improved, complete recovery of anaphora is ensured, and the intellectualization capability of a dialogue system is improved.