CN111126068A - Chinese named entity recognition method and device and electronic equipment - Google Patents
Chinese named entity recognition method and device and electronic equipment Download PDFInfo
- Publication number
- CN111126068A CN111126068A CN201911360655.9A CN201911360655A CN111126068A CN 111126068 A CN111126068 A CN 111126068A CN 201911360655 A CN201911360655 A CN 201911360655A CN 111126068 A CN111126068 A CN 111126068A
- Authority
- CN
- China
- Prior art keywords
- text
- language understanding
- model
- training
- named entity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000012549 training Methods 0.000 claims abstract description 125
- 230000002457 bidirectional effect Effects 0.000 claims abstract description 65
- 230000011218 segmentation Effects 0.000 claims abstract description 56
- 238000013528 artificial neural network Methods 0.000 claims abstract description 53
- 238000013135 deep learning Methods 0.000 claims abstract description 40
- 238000002372 labelling Methods 0.000 claims abstract description 26
- 239000013598 vector Substances 0.000 claims description 54
- 238000003062 neural network model Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 10
- 230000015654 memory Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 description 19
- 238000010586 diagram Methods 0.000 description 5
- 239000003814 drug Substances 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000013519 translation Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Machine Translation (AREA)
Abstract
The invention provides a method and a device for identifying a Chinese named entity and electronic equipment, wherein the method comprises the following steps: acquiring a predicted text for identifying a named entity; inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model; and identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities. The invention can train the multilayer bidirectional neural network through a deep learning algorithm to obtain a language understanding model, and solves the problems of low accuracy, incapability of realizing word ambiguity, incapability of predicting out-of-vocabulary word segmentation and low training speed of the conventional named entity recognition method.
Description
Technical Field
The invention relates to the technical field of language models, in particular to a method and a device for identifying a Chinese named entity and electronic equipment.
Background
Named entity recognition refers to a process of recognizing a name or symbol of a thing of a specific entity type such as a name of a person, a place, a name of an organization, a movie, a novel, a game, or a name of a medicine in text information, and can be applied to fields of information retrieval, a question-answering system, syntactic analysis, machine translation, and the like. For example, in the medical industry, drug names can be accurately identified in a piece of text by named entity identification; or, in the machine translation field, the terminal firstly carries out named entity recognition on the input text information and then translates the text information according to the recognized named entity, so that the accuracy of machine translation is improved.
At present, one way of identifying a named entity is to pre-train a word2vec model to enable a machine to understand word segmentation in a vocabulary, in a specific pre-training process, obtain a large amount of original text data, perform word segmentation on the text data by using a word segmentation algorithm, or directly obtain a large amount of original word segmentation, input the word segmentation into the word2vec model for pre-training, and obtain a word vector of the word segmentation, so that after the pre-training is completed, the word vector of the word segmentation in the vocabulary can be identified through the word2vec model, and thus the word segmentation in the vocabulary can be understood.
In the named entity recognition process, obtaining a text recognized by the standby name entity, carrying out word segmentation on the text by using a word segmentation algorithm, inputting the obtained word segmentation into a word2vec model, and understanding which word belongs to the word in a vocabulary table by using a word vector output by the word2vec model; combining the obtained vectors of all the participles to obtain a vector of the text; inputting the obtained text vector into a long-term memory network LSTM to carry out text semantic understanding (also forming a coding process), and expressing the understood semantic meaning by using a corresponding vector; and carrying out named entity labeling (also called a decoding process) by using a conditional random field CRF according to the semantic meaning corresponding to the vector to obtain a named entity labeling result.
The scheme of understanding word segmentation and carrying out named entity labeling by using the word2vec model has the following problems:
1) the method includes the steps that original text data need to be segmented through a word segmentation algorithm, but the accuracy of the existing Chinese word segmentation algorithm is unstable, and the result of word segmentation can have great influence on the result of a word2vec word vector prediction model;
2) after word2vec model is pre-trained, in the process of naming entity labeling, word vectors corresponding to words in a vocabulary table in a text can be obtained, but the same word segmentation only obtains a unique vector, and the change cannot be carried out according to the meaning of the context of the input text, so that the word ambiguity cannot be realized;
3) in the word2vec model pre-training, in order to ensure that a machine can understand texts, a large number of words are required to be used for pre-training, so that the training speed of the word2vec word vector prediction model is low, and after the word2vec model pre-training is finished, only the word segmentation in the word list can be understood, and the word vectors corresponding to the word segmentation outside the word list cannot be predicted.
Disclosure of Invention
The invention provides a method and a device for identifying a Chinese named entity and electronic equipment, which are used for solving the problems that the accuracy is not high, the word ambiguity cannot be realized, the word segmentation outside a vocabulary cannot be predicted and the training speed is slow in the scheme of understanding the word segmentation by using a word2vec model and labeling the named entity in the prior art.
In a first aspect, the present application provides a method for identifying a named entity in chinese, including:
acquiring a predicted text for identifying a named entity;
inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities.
Optionally, the language understanding model is a multi-layer bidirectional neural network model that represents BER based on bidirectional coding.
Optionally, training a multi-layer neural network in advance based on a deep learning algorithm by using a plurality of original texts as training samples to obtain the language understanding model, including:
using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, and determining model parameters of the multilayer neural network to obtain the language understanding pre-training model;
the method comprises the steps of using a plurality of original texts with named entity labels as training samples, training a language understanding pre-training model by adopting a supervised learning algorithm, and adjusting model parameters of the language understanding pre-training model to obtain the language understanding model.
Optionally, identifying the participles belonging to the named entity in the text based on the text semantic representation information, and performing named entity tagging, including:
based on the text semantic representation information, coding by using a long-time memory network (LSTM) and correcting the text semantic representation information;
and decoding the text semantic representation information by using a conditional random field CRF, identifying the participles belonging to the named entities in the text, and labeling the named entities.
Optionally, before inputting the predicted text into a language understanding model or before inputting the original text into a multi-layer bidirectional neural network, the method further includes:
and carrying out position coding on the position of each single character according to the position of each single character in the predicted text or the original text.
Optionally, training a multi-layer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model, including:
and training the multilayer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model.
Optionally, the text semantic representation information obtained by performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model includes:
and adopting a text vector to represent the text semantic representation information, wherein the text vector is a combination of a plurality of word vectors in the text.
In a second aspect, the present application provides a device for identifying a named entity in chinese, the device comprising:
the acquisition module is used for acquiring a prediction text for identifying the named entity;
the prediction module is used for inputting the predicted text into a language understanding model and performing word segmentation bidirectional prediction on the text according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and the marking module is used for identifying the participles belonging to the named entities in the text based on the text semantic representation information and marking the named entities.
Optionally, the language understanding model is a multi-layer bidirectional neural network model that represents BER based on bidirectional coding.
Optionally, the prediction module is configured to use a plurality of original texts as training samples, train a multi-layer neural network in advance based on a deep learning algorithm to obtain the language understanding model, and specifically configured to:
using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, and determining model parameters of the multilayer neural network to obtain the language understanding pre-training model;
the method comprises the steps of using a plurality of original texts with named entity labels as training samples, training a language understanding pre-training model by adopting a supervised learning algorithm, and adjusting model parameters of the language understanding pre-training model to obtain the language understanding model.
Optionally, the labeling module is configured to identify a participle belonging to a named entity in the text based on the text semantic representation information, and perform named entity labeling, specifically configured to:
based on the text semantic representation information, coding by using a long-time memory network (LSTM) and correcting the text semantic representation information;
and decoding the text semantic representation information by using a conditional random field CRF, identifying the participles belonging to the named entities in the text, and labeling the named entities.
Optionally, the predicting module is configured to, before inputting the predicted text into the language understanding model or before inputting the original text into the multi-layer bidirectional neural network, further include:
and the position coding module is used for carrying out position coding on the position of each single character according to the position of each single character in the predicted text or the original text.
Optionally, the prediction model is used for training a multi-layer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model, and is specifically used for:
and training the multilayer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model.
Optionally, the prediction module is configured to perform word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, and is specifically configured to:
and adopting a text vector to represent the text semantic representation information, wherein the text vector is a combination of a plurality of word vectors in the text.
In a third aspect, the present application provides an electronic device for chinese named entity recognition, comprising: a storage unit and a processing unit;
wherein the storage unit is used for storing programs;
the processing unit is used for executing the program in the storage unit and comprises the following steps:
acquiring a predicted text for identifying a named entity;
inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities.
In a fourth aspect, the present application also provides a computer storage medium having a computer program stored thereon, which when executed by a processing unit, performs the steps of the method of the first aspect.
In a fifth aspect, the present application further provides a computer program product, including a computer program, where the computer program includes program instructions, and when the program instructions are executed by an electronic device, the electronic device is caused to execute the audio/video preloading method of any one of the above items.
In addition, for technical effects brought by any one implementation manner of the second aspect to the fifth aspect, reference may be made to technical effects brought by different implementation manners of the first aspect, and details are not described here.
The method and the device for identifying the Chinese named entity and the electronic equipment have the following beneficial effects that:
according to the method and the device for recognizing the Chinese named entity and the electronic equipment, the multilayer bidirectional neural network model can be trained based on the deep learning algorithm to obtain the language understanding model provided by the application, the language understanding model obtained by training can be better combined with the semantic representation information of the context prediction text by adopting the double-layer neural network model, and the semantic representation information obtained by adopting the multilayer neural network model is more accurate, so that the Chinese named entity in the text can be better recognized and labeled.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
FIG. 1 is a diagram illustrating a method for naming an entity in Chinese according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating a method for naming an entity in Chinese according to an embodiment of the present invention;
FIG. 3 is a diagram of a Chinese named entity apparatus according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of an electronic device for naming a Chinese entity according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail with reference to the accompanying drawings, and it is apparent that the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The term "and/or" in the embodiments of the present invention describes an association relationship of associated objects, and indicates that three relationships may exist, for example, a and/or B may indicate: a exists alone, A and B exist simultaneously, and B exists alone. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship.
For convenience of understanding, terms referred to in the embodiments of the present invention are explained below:
and (3) deep learning algorithm: deep learning is to learn the internal rules and the expression levels of sample data, and the final aim is to enable a machine to have the analysis and learning capability like a human and to recognize data such as characters, images and sounds.
The application scenario described in the embodiment of the present invention is for more clearly illustrating the technical solution of the embodiment of the present invention, and does not form a limitation on the technical solution provided in the embodiment of the present invention, and it can be known by a person skilled in the art that with the occurrence of a new application scenario, the technical solution provided in the embodiment of the present invention is also applicable to similar technical problems. In the description of the present invention, the term "plurality" means two or more unless otherwise specified.
Named entity recognition refers to recognition of entities with specific meanings in text, such as proper nouns like personal names, organization names, place names, and meaningful time, and specific categories can appear in specific fields in some special fields, such as medicine names, diseases, and the like, and is a basic task of technologies like information retrieval and question-answering systems. As in "xiaoming on hawaii vacation", named entities are: compared with English named entity recognition, the Chinese named entity has no obvious form mark and word segmentation interference, so that the Chinese named entity recognition difficulty is higher than that of English.
The existing named entity recognition method is characterized in that a large amount of original text data are segmented by using a word segmentation algorithm, pre-training is carried out according to a word2vec model to obtain word vectors of the segmented words, and the word vectors are vectors capable of converting characters in real world abstraction into vectors capable of carrying out mathematical formula operation, so that after the pre-training is completed, the word vectors of the segmented words in a vocabulary table can be recognized through the word2vec model, and the segmented words in the vocabulary table can be understood. However, as mentioned above, the word segmentation difficulty of the text in the chinese language is high, and there is a certain inaccuracy due to the word segmentation algorithm, so that the accuracy of the word vector is affected, and the result of word segmentation has a large influence on the result of the word2vec word vector prediction model; after word2vec model is pre-trained, in the process of naming entity labeling, word vectors corresponding to words in a vocabulary table in a text can be obtained, but the same word segmentation only obtains a unique vector, and the change cannot be carried out according to the meaning of the context of the input text, so that the word ambiguity cannot be realized; in the word2vec model pre-training, in order to ensure that a machine can understand a text, a large number of words are required to be used for pre-training, so that the training speed of a word2vec word vector prediction model is low, after the word2vec model pre-training is finished, only the word segmentation in a word list can be understood, the word vector corresponding to the word segmentation outside the word list cannot be predicted, the word order of the input text cannot be considered in the training process of the word2vec word vector prediction model, and the word vector cannot be accurately obtained according to the word order/semantic information of the text.
Based on the above-mentioned problems, the present application provides a method for identifying named entities in chinese, the present application trains a multi-layer bidirectional neural network model to obtain a language understanding model based on a deep learning algorithm, wherein, a plurality of original texts without named entity labels are used as training samples, the multi-layer neural network is trained in advance based on the deep learning algorithm, model parameters of the multi-layer neural network are determined to obtain the language understanding pre-training model, the obtained language understanding pre-training model has the capability of obtaining text semantic representation information of a predicted text, but a plurality of original texts with named entity labels are further used as training samples, a supervised learning algorithm is used to train the language understanding pre-training model, model parameters of the language understanding pre-training model are adjusted to obtain the language understanding model, the language understanding model at this time can accurately predict the predicted text of the input model.
As shown in fig. 1, a method for identifying a chinese named entity provided in an embodiment of the present application includes:
step S101, acquiring a prediction text for identifying a named entity;
the predicted text is a text needing to be predicted to obtain text semantic representation information, and has different categories in different fields, for example, in the medical field, after a medical history of a patient is input, the text describing the medical history needs to be understood, symptoms and medicine names need to be accurately found out from the text, in the intelligent voice field, when a user needs to understand the meaning of a user utterance, and key named entities need to be found out, and as mentioned above, the text needing to understand the semantics can be obtained in different application fields and can be used as the predicted text;
and obtaining a prediction text for identifying the named entity, and before inputting the prediction text into the reason understanding model, carrying out position coding on the position of each single character according to the position of each single character in the prediction text to obtain the position representation of each single character in the text.
Step S102, inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single characters in the text and the context of each single character by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
the language understanding model obtained in the embodiment of the application can predict the predicted text of the input language understanding model to obtain text semantic representation information, wherein a text vector is used for representing the text semantic representation information, and the text vector is a combination of a plurality of word vectors in the text;
the embodiment of the present application is based on deep learning, a language understanding model is obtained by training a multi-layer bidirectional Neural network, a language understanding model with a good effect can be obtained in a training process by using the multi-layer bidirectional Neural network, that is, the Neural network of each layer is bidirectional, the embodiment of the present application uses a multi-layer bidirectional Neural network model based on BER (bidirectional encoder Representation), the multi-layer bidirectional Neural network model may include, but is not limited to, a transform model, CNN (Convolutional Neural Networks) model or other Neural network models, the transform is an attention mechanism, and can learn context relationships between words in a text, wherein the BER-based transform model is a BERT (bidirectional encoder Representation) model, the training process is divided into a pre-training process and a retraining process, and the BERT model is taken as an example below, the training process is specifically described as follows:
1) the pre-training process comprises the following steps: using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, and determining model parameters of the multilayer neural network to obtain the language understanding pre-training model;
and training the Transformer multilayer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding pre-training model.
The method comprises the steps of inputting a plurality of original texts without named entity labels into a BERT model as training samples, determining model parameters of the obtained language understanding pre-training model through an unsupervised learning training mode, wherein the original texts without named entity labels directly use the obtained original texts as the training samples, the original texts without named entity labels are an original text set with a large number, and as an optional implementation mode, the original texts can be directly used as the training samples, or the original texts are divided into a plurality of groups of original samples with different categories according to different fields, and the original samples without named entity labels of each group are respectively input into the BERT model to obtain a plurality of language understanding pre-training models with different functions.
The language understanding pre-training model can only simply predict the semantic representation information of the text, and the language understanding model obtained through the following re-training process is required to accurately predict the semantic representation information of the text.
2) And (3) retraining: the method comprises the steps of using a plurality of original texts with named entity labels as training samples, training a language understanding pre-training model by adopting a supervised learning algorithm, and adjusting model parameters of the language understanding pre-training model to obtain the language understanding model.
Inputting a plurality of original texts with named entity labels as training samples into a language understanding pre-training model obtained by the pre-training, adjusting parameters of the language understanding pre-training model by a training mode with supervised learning to obtain a language understanding model, and correctly labeling all named entities in the original texts in advance to obtain the original texts with the named entity labels;
as mentioned above, the original texts with named entity labels may be the same as or different from the original texts without named entity labels, and may be a large number of original text sets, optionally, a plurality of original texts may be used as training samples, or, the original texts may be divided into a plurality of groups of original texts with different categories according to different fields, and the groups of original texts with named entity labels are respectively input into the one or more language understanding pre-training models with different functions, so as to obtain one language understanding model or a plurality of language understanding models with different functions.
And carrying out position coding on the position of each single character according to the position of each single character in the original text without the named entity label or the position of each single character in the original text with the named entity label to obtain the position representation of each single character in the original text.
And S103, identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities.
Based on the text semantic representation information of the predicted text obtained through the language understanding model, the text semantic representation information can be directly decoded by using a Conditional Random Field (CRF), the participles belonging to named entities in the text are identified, and the named entities are labeled;
as an optional implementation manner, firstly, based on the text semantic representation information, a Long Short-Term Memory (LSTM) is used for encoding, the text semantic representation information is corrected, then, CRF is used for decoding the text semantic representation information, and the participles belonging to the named entity in the text are identified, and the named entity is labeled.
The above-mentioned determining the text semantic representation information respectively includes but is not limited to the following two ways:
1) determining a text vector of a predicted text as text semantic representation information
After the predicted text is input into a language understanding model, firstly, the predicted text is divided into single characters, word segmentation is carried out according to the context of each single character, each single character is combined with at least one single character in the text, word segmentation is obtained, the meaning of the predicted text is predicted according to the meaning of the obtained word segmentation or the combination of the obtained word vectors is obtained, text vectors are obtained, and text semantic representation information is represented in the form of the text vectors, wherein the text vectors can be multi-dimensional text vectors, and are not limited;
the language understanding model reads the whole text at one time, but does not read the text word segmentation in the prior art, or reads the whole text in a left-to-right sequence, or reads the whole text in a right-to-left sequence, can learn based on two sides of the text, is a bidirectional prediction language understanding model, and understands the text context more deeply than a unidirectional language understanding model.
2) Determining word vectors of each participle of predicted text as text semantic representation information
After the predicted text is input into a language understanding model, firstly, the predicted text is divided into single characters, word segmentation is carried out according to the context of each single character, each single character is combined with at least one single character in the text, and each single character is combined with at least one single character in the text and at least one single character in the text to obtain word segmentation, and word vectors of the word segmentation are used as text semantic representation information of the predicted text, wherein the word vectors of the word segmentation can be word vectors with the same dimension, and no limitation is made here.
The method for obtaining the single words according to the predicted text and obtaining the word segmentation based on the context avoids the problem that the inaccuracy of the semantic expression mode of the predicted text is influenced by the inaccuracy of analysis caused by the fact that word segmentation is required to be performed by a word segmentation algorithm in advance.
The BERT model is a language representation model, a deep bidirectional neural network model is pre-trained on the basis of left and right contexts of all layers of a neural network, then fine tuning is carried out by using an additional output layer, and a better neural network model is further created for downstream tasks, such as question answering and language inference tasks, and the architecture of the neural network does not need to be greatly modified aiming at specific tasks in the fine tuning process.
At present, strategies for establishing a better neural network model for a downstream task comprise fine tuning and the like, the least parameters corresponding to a specific task are introduced, only the parameters obtained by pre-training are subjected to fine tuning, and the parameters are trained in the downstream task to obtain a final neural network model.
The BERT model proposes a new pre-training mode including a Masked Language Model (MLM) that randomly masks a part of words input to a multi-layer bidirectional neural network model, with the goal of predicting the Masked words based on left, right, and context, and randomly masks 15% of the words as training samples, wherein 80% of the Masked part is replaced with Masked token, 10% is replaced with a random word, and 10% is left unchanged.
The following describes in detail a method for identifying a named entity in chinese, which is proposed by the present application, with reference to a specific embodiment, as shown in fig. 2, including:
step S201, using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, determining model parameters of the multilayer neural network to obtain a language understanding pre-training model;
step S202, a plurality of original texts with named entity labels are used as training samples, a supervised learning algorithm is adopted to train a language understanding pre-training model, and model parameters of the language understanding pre-training model are adjusted to obtain a language understanding model;
step S203, acquiring a prediction text for identifying the named entity;
step S204, inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single characters in the text and the context of each single character by using the language understanding model to obtain text semantic representation information;
step S205, based on the text semantic representation information, using a long-time memory network LSTM to encode, and correcting the text semantic representation information;
and S206, decoding the text semantic representation information by using a conditional random field CRF, identifying the participles belonging to the named entities in the text, and labeling the named entities.
The above description is directed to a method for identifying a named entity in Chinese, and the following description is directed to a device for performing the named entity identification.
Please refer to fig. 3, which illustrates an apparatus for recognizing a chinese named entity according to an embodiment of the present invention, including:
an obtaining module 301, configured to obtain a prediction text for identifying a named entity;
the prediction module 302 is configured to input the predicted text into a language understanding model, and perform word segmentation and bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, where a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and the labeling module 303 is configured to identify a participle belonging to a named entity in the text based on the text semantic representation information, and perform named entity labeling.
Optionally, the language understanding model is a multi-layer bidirectional neural network model that represents BER based on bidirectional coding.
Optionally, the prediction module is configured to use a plurality of original texts as training samples, train a multi-layer neural network in advance based on a deep learning algorithm to obtain the language understanding model, and specifically configured to:
using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, and determining model parameters of the multilayer neural network to obtain the language understanding pre-training model;
the method comprises the steps of using a plurality of original texts with named entity labels as training samples, training a language understanding pre-training model by adopting a supervised learning algorithm, and adjusting model parameters of the language understanding pre-training model to obtain the language understanding model.
Optionally, the labeling module is configured to identify a participle belonging to a named entity in the text based on the text semantic representation information, and perform named entity labeling, specifically configured to:
based on the text semantic representation information, coding by using a long-time memory network (LSTM) and correcting the text semantic representation information;
and decoding the text semantic representation information by using a conditional random field CRF, identifying the participles belonging to the named entities in the text, and labeling the named entities.
Optionally, the predicting module is configured to, before inputting the predicted text into the language understanding model or before inputting the original text into the multi-layer bidirectional neural network, further include:
and the position coding module 304 is configured to perform position coding on the position of each single character according to the position of each single character in the predicted text or the original text.
Optionally, the prediction model is used for training a multi-layer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model, and is specifically used for:
and training the multilayer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model.
Optionally, the prediction module is configured to perform word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, and is specifically configured to:
and adopting a text vector to represent the text semantic representation information, wherein the text vector is a combination of a plurality of word vectors in the text.
The above describes a chinese named entity recognition apparatus in the embodiment of the present application from the perspective of a modular functional entity, and the following describes an electronic device for chinese named entity recognition in the embodiment of the present application from the perspective of hardware processing.
Referring to fig. 4, an electronic device for recognizing a named entity in chinese in an embodiment of the present application includes:
at least one processing unit 401 and at least one memory unit 402, and a bus system 409;
wherein the storage unit stores program code that, when executed by the processing unit, causes the processing unit to perform the following:
acquiring a predicted text for identifying a named entity;
inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities.
Fig. 4 is a schematic diagram of an electronic device for chinese named entity recognition according to an embodiment of the present disclosure, where the device 400 may have a relatively large difference due to different configurations or performances, and may include one or more processing units (CPU) 401 (e.g., one or more processing units) and a storage unit 402, one or more storage media 403 (e.g., one or more mass storage devices) for storing an application program 404 or data 405. The storage unit 402 and the storage medium 403 may be a transient storage or a persistent storage, among others. The program stored in the storage medium 403 may include one or more modules (not shown), and each module may include a series of instruction operations in the information processing apparatus. Still further, the processing unit 401 may be arranged to communicate with the storage medium 403, executing a series of instruction operations in the storage medium 403 on the device 400.
The device 400 may also include one or more wired or wireless network interfaces 407, one or more input-output interfaces 408, and/or one or more operating systems 406, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc.
Optionally, the language understanding model is a multi-layer bidirectional neural network model that represents BER based on bidirectional coding.
Optionally, training a multi-layer neural network in advance based on a deep learning algorithm by using a plurality of original texts as training samples to obtain the language understanding model, including:
using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, and determining model parameters of the multilayer neural network to obtain the language understanding pre-training model;
the method comprises the steps of using a plurality of original texts with named entity labels as training samples, training a language understanding pre-training model by adopting a supervised learning algorithm, and adjusting model parameters of the language understanding pre-training model to obtain the language understanding model.
Optionally, identifying the participles belonging to the named entity in the text based on the text semantic representation information, and performing named entity tagging, including:
based on the text semantic representation information, coding by using a long-time memory network (LSTM) and correcting the text semantic representation information;
and decoding the text semantic representation information by using a conditional random field CRF, identifying the participles belonging to the named entities in the text, and labeling the named entities.
Optionally, before inputting the predicted text into a language understanding model or before inputting the original text into a multi-layer bidirectional neural network, the method further includes:
and carrying out position coding on the position of each single character according to the position of each single character in the predicted text or the original text.
Optionally, training a multi-layer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model, including:
and training the multilayer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model.
Optionally, the processing unit is configured to perform word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, and includes:
and adopting a text vector to represent the text semantic representation information, wherein the text vector is a combination of a plurality of word vectors in the text.
The embodiment of the invention also provides a computer-readable storage medium, which comprises instructions, and when the computer-readable storage medium runs on a computer, the computer is enabled to execute the method for identifying the Chinese named entity provided by the embodiment.
An embodiment of the present application further provides a computer program product, which includes a computer program, where the computer program includes program instructions, and when the program instructions are executed by an electronic device, the electronic device is enabled to execute the method for identifying a named entity in chinese provided in the foregoing embodiment.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses and modules may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and in actual implementation, there may be other divisions, for example, multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or modules, and may be in an electrical, mechanical or other form.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical modules, may be located in one place, or may be distributed on a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present application may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may be stored in a computer readable storage medium.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product.
The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the application to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can store or a data storage device, such as a server, a data center, etc., that is integrated with one or more available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
The technical solutions provided by the present application are introduced in detail, and the present application applies specific examples to explain the principles and embodiments of the present application, and the descriptions of the above examples are only used to help understand the method and the core ideas of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.
Claims (10)
1. A Chinese named entity recognition method is characterized by comprising the following steps:
acquiring a predicted text for identifying a named entity;
inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities.
2. The method of claim 1, wherein the language understanding model is a multi-layer bidirectional neural network model that represents BER using bidirectional coding.
3. The method according to claim 1 or 2, wherein the training of the multi-layer neural network based on a deep learning algorithm in advance to obtain the language understanding model by using a plurality of original texts as training samples comprises:
using a plurality of original texts without named entity labels as training samples, training a multilayer neural network in advance based on a deep learning algorithm, and determining model parameters of the multilayer neural network to obtain the language understanding pre-training model;
the method comprises the steps of using a plurality of original texts with named entity labels as training samples, training a language understanding pre-training model by adopting a supervised learning algorithm, and adjusting model parameters of the language understanding pre-training model to obtain the language understanding model.
4. The method according to claim 1 or 2, wherein identifying the participles belonging to the named entity in the text based on the text semantic representation information, and performing named entity tagging comprises:
based on the text semantic representation information, coding by using a long-time memory network (LSTM) and correcting the text semantic representation information;
and decoding the text semantic representation information by using a conditional random field CRF, identifying the participles belonging to the named entities in the text, and labeling the named entities.
5. The method of claim 1, wherein prior to inputting the predicted text into a language understanding model or prior to inputting the original text into a multi-layered bidirectional neural network, further comprising:
and carrying out position coding on the position of each character according to the position of each single character in the predicted text or the original text.
6. The method according to claim 1 or 2, wherein the training of the multi-layer bidirectional neural network based on a deep learning algorithm in advance to obtain the language understanding model comprises:
and training the multilayer bidirectional neural network in advance based on a deep learning algorithm to obtain the language understanding model.
7. The method of claim 1, wherein the semantic representation information of the text obtained by performing word segmentation and bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model comprises:
and adopting a text vector to represent the text semantic representation information, wherein the text vector is a combination of a plurality of word vectors in the text.
8. A chinese named entity recognition device, comprising:
the acquisition module is used for acquiring a prediction text for identifying the named entity;
the prediction module is used for inputting the predicted text into a language understanding model and performing word segmentation bidirectional prediction on the text according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and the marking module is used for identifying the participles belonging to the named entities in the text based on the text semantic representation information and marking the named entities.
9. An electronic device for Chinese named entity recognition, comprising: a storage unit and a processing unit;
wherein the storage unit is used for storing programs;
the processing unit is used for executing the program in the storage unit and comprises the following steps:
acquiring a predicted text for identifying a named entity;
inputting the predicted text into a language understanding model, and performing word segmentation bidirectional prediction according to the single words in the text and the context of each single word by using the language understanding model to obtain text semantic representation information, wherein a plurality of original texts are used as training samples, and a multi-layer bidirectional neural network is trained in advance based on a deep learning algorithm to obtain the language understanding model;
and identifying the participles belonging to the named entities in the text based on the text semantic representation information, and labeling the named entities.
10. A computer-readable storage medium comprising computer program instructions which, when run on a computer, cause the computer to perform the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911360655.9A CN111126068A (en) | 2019-12-25 | 2019-12-25 | Chinese named entity recognition method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911360655.9A CN111126068A (en) | 2019-12-25 | 2019-12-25 | Chinese named entity recognition method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111126068A true CN111126068A (en) | 2020-05-08 |
Family
ID=70502541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911360655.9A Pending CN111126068A (en) | 2019-12-25 | 2019-12-25 | Chinese named entity recognition method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111126068A (en) |
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111444721A (en) * | 2020-05-27 | 2020-07-24 | 南京大学 | Chinese text key information extraction method based on pre-training language model |
| CN111708878A (en) * | 2020-08-20 | 2020-09-25 | 科大讯飞(苏州)科技有限公司 | Method, device, storage medium and equipment for extracting sports text abstract |
| CN111859967A (en) * | 2020-06-12 | 2020-10-30 | 北京三快在线科技有限公司 | Entity identification method and device and electronic equipment |
| CN111986765A (en) * | 2020-09-03 | 2020-11-24 | 平安国际智慧城市科技股份有限公司 | Electronic case entity marking method, device, computer equipment and storage medium |
| CN111985239A (en) * | 2020-07-31 | 2020-11-24 | 杭州远传新业科技有限公司 | Entity identification method and device, electronic equipment and storage medium |
| CN112101027A (en) * | 2020-07-24 | 2020-12-18 | 昆明理工大学 | Chinese named entity recognition method based on reading understanding |
| CN112101032A (en) * | 2020-08-31 | 2020-12-18 | 广州探迹科技有限公司 | Named entity identification and error correction method based on self-distillation |
| CN112287683A (en) * | 2020-08-19 | 2021-01-29 | 北京沃东天骏信息技术有限公司 | Named entity identification method and device |
| CN112347773A (en) * | 2020-10-26 | 2021-02-09 | 北京诺道认知医学科技有限公司 | Medical application model training method and device based on BERT model |
| CN112765984A (en) * | 2020-12-31 | 2021-05-07 | 平安资产管理有限责任公司 | Named entity recognition method and device, computer equipment and storage medium |
| CN112801085A (en) * | 2021-02-09 | 2021-05-14 | 沈阳麟龙科技股份有限公司 | Method, device, medium and electronic equipment for recognizing characters in image |
| CN112861515A (en) * | 2021-02-08 | 2021-05-28 | 上海天壤智能科技有限公司 | Interactive knowledge definition and processing method, system, device and readable medium |
| CN112906381A (en) * | 2021-02-02 | 2021-06-04 | 北京有竹居网络技术有限公司 | Recognition method and device of conversation affiliation, readable medium and electronic equipment |
| CN112966507A (en) * | 2021-03-29 | 2021-06-15 | 北京金山云网络技术有限公司 | Method, device, equipment and storage medium for constructing recognition model and identifying attack |
| CN112989829A (en) * | 2021-02-10 | 2021-06-18 | 海尔数字科技（上海）有限公司 | Named entity identification method, device, equipment and storage medium |
| CN113051875A (en) * | 2021-03-22 | 2021-06-29 | 北京百度网讯科技有限公司 | Training method of information conversion model, and text information conversion method and device |
| CN113051913A (en) * | 2021-04-09 | 2021-06-29 | 中译语通科技股份有限公司 | Tibetan word segmentation information processing method, system, storage medium, terminal and application |
| CN113139069A (en) * | 2021-05-14 | 2021-07-20 | 上海交通大学 | Knowledge graph construction-oriented Chinese text entity identification method and system for power failure |
| CN113157883A (en) * | 2021-04-07 | 2021-07-23 | 浙江工贸职业技术学院 | Chinese opinion target boundary prediction method based on dual-model structure |
| CN113204969A (en) * | 2021-05-31 | 2021-08-03 | 平安科技（深圳）有限公司 | Medical named entity recognition model generation method and device and computer equipment |
| WO2021151322A1 (en) * | 2020-09-07 | 2021-08-05 | 平安科技（深圳）有限公司 | Method and apparatus for entity identification based on deep learning model, device, and medium |
| CN113221565A (en) * | 2021-05-07 | 2021-08-06 | 北京百度网讯科技有限公司 | Entity recognition model training method and device, electronic equipment and storage medium |
| CN113408619A (en) * | 2021-06-21 | 2021-09-17 | 江苏苏云信息科技有限公司 | Language model pre-training method and device |
| CN113420822A (en) * | 2021-06-30 | 2021-09-21 | 北京百度网讯科技有限公司 | Model training method and device and text prediction method and device |
| CN113673247A (en) * | 2021-05-13 | 2021-11-19 | 江苏曼荼罗软件股份有限公司 | Entity identification method, device, medium and electronic equipment based on deep learning |
| CN113807094A (en) * | 2020-06-11 | 2021-12-17 | 株式会社理光 | Entity identification method, device and computer readable storage medium |
| CN113849624A (en) * | 2021-10-15 | 2021-12-28 | 上海大参林医疗健康科技有限公司 | Word slot extraction device and method for multi-turn conversation |
| CN114036956A (en) * | 2021-11-18 | 2022-02-11 | 清华大学 | Tourism knowledge semantic analysis method and device |
| CN114118093A (en) * | 2022-01-27 | 2022-03-01 | 华东交通大学 | Method and system for identifying flat mark enhanced nested named entity |
| CN114385809A (en) * | 2020-10-22 | 2022-04-22 | 中移(成都)信息通信科技有限公司 | Training method, device and equipment of entity text extraction model |
| CN114492383A (en) * | 2021-12-20 | 2022-05-13 | 北京邮电大学 | Entity name identification method and device for digital currency transaction address |
| CN114564971A (en) * | 2022-02-28 | 2022-05-31 | 北京百度网讯科技有限公司 | Deep learning model training method, text data processing method and text data processing device |
| CN114648113A (en) * | 2022-03-14 | 2022-06-21 | 中移(苏州)软件技术有限公司 | Named entity recognition neural structure in reading understanding form |
| CN114706942A (en) * | 2022-03-16 | 2022-07-05 | 马上消费金融股份有限公司 | Text conversion model training method, text conversion device and electronic equipment |
| CN114707513A (en) * | 2022-03-22 | 2022-07-05 | 腾讯科技（深圳）有限公司 | Text semantic recognition method and device, electronic equipment and storage medium |
| CN116029299A (en) * | 2023-03-30 | 2023-04-28 | 北京融信数联科技有限公司 | Named entity recognition method, system and storage medium based on polysemous words |
| CN116070629A (en) * | 2023-04-06 | 2023-05-05 | 北京蜜度信息技术有限公司 | Chinese text word order checking method, system, storage medium and electronic equipment |
| CN118133830A (en) * | 2024-04-30 | 2024-06-04 | 北京壹永科技有限公司 | Named entity recognition method, named entity recognition device, named entity recognition equipment and named entity recognition computer readable storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106601228A (en) * | 2016-12-09 | 2017-04-26 | 百度在线网络技术（北京）有限公司 | Sample marking method and device based on artificial intelligence prosody prediction |
| CN109359293A (en) * | 2018-09-13 | 2019-02-19 | 内蒙古大学 | Mongolian name entity recognition method neural network based and its identifying system |
| CN109635279A (en) * | 2018-11-22 | 2019-04-16 | 桂林电子科技大学 | A kind of Chinese name entity recognition method neural network based |
| CN109933801A (en) * | 2019-03-25 | 2019-06-25 | 北京理工大学 | Two-way LSTM based on predicted position attention names entity recognition method |
| CN109979435A (en) * | 2017-12-28 | 2019-07-05 | 北京搜狗科技发展有限公司 | Data processing method and device, the device for data processing |
| CN110010217A (en) * | 2019-04-11 | 2019-07-12 | 中国医学科学院医学信息研究所 | A kind of mask method and device of electronic health record |
| CN110083831A (en) * | 2019-04-16 | 2019-08-02 | 武汉大学 | A kind of Chinese name entity recognition method based on BERT-BiGRU-CRF |
| CN110287494A (en) * | 2019-07-01 | 2019-09-27 | 济南浪潮高新科技投资发展有限公司 | A method of the short text Similarity matching based on deep learning BERT algorithm |
-
2019
- 2019-12-25 CN CN201911360655.9A patent/CN111126068A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106601228A (en) * | 2016-12-09 | 2017-04-26 | 百度在线网络技术（北京）有限公司 | Sample marking method and device based on artificial intelligence prosody prediction |
| CN109979435A (en) * | 2017-12-28 | 2019-07-05 | 北京搜狗科技发展有限公司 | Data processing method and device, the device for data processing |
| CN109359293A (en) * | 2018-09-13 | 2019-02-19 | 内蒙古大学 | Mongolian name entity recognition method neural network based and its identifying system |
| CN109635279A (en) * | 2018-11-22 | 2019-04-16 | 桂林电子科技大学 | A kind of Chinese name entity recognition method neural network based |
| CN109933801A (en) * | 2019-03-25 | 2019-06-25 | 北京理工大学 | Two-way LSTM based on predicted position attention names entity recognition method |
| CN110010217A (en) * | 2019-04-11 | 2019-07-12 | 中国医学科学院医学信息研究所 | A kind of mask method and device of electronic health record |
| CN110083831A (en) * | 2019-04-16 | 2019-08-02 | 武汉大学 | A kind of Chinese name entity recognition method based on BERT-BiGRU-CRF |
| CN110287494A (en) * | 2019-07-01 | 2019-09-27 | 济南浪潮高新科技投资发展有限公司 | A method of the short text Similarity matching based on deep learning BERT algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 丁龙，文雯，林强: "基于预训练BERT字嵌入模型的领域实体识别", 《情报工程》 * |
| 丁龙，文雯，林强: "基于预训练BERT字嵌入模型的领域实体识别", 《情报工程》, vol. 5, no. 6, 15 December 2019 (2019-12-15), pages 65 - 73 * |
Cited By (50)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111444721A (en) * | 2020-05-27 | 2020-07-24 | 南京大学 | Chinese text key information extraction method based on pre-training language model |
| CN113807094B (en) * | 2020-06-11 | 2024-03-19 | 株式会社理光 | Entity recognition method, entity recognition device and computer readable storage medium |
| CN113807094A (en) * | 2020-06-11 | 2021-12-17 | 株式会社理光 | Entity identification method, device and computer readable storage medium |
| CN111859967A (en) * | 2020-06-12 | 2020-10-30 | 北京三快在线科技有限公司 | Entity identification method and device and electronic equipment |
| CN111859967B (en) * | 2020-06-12 | 2024-04-09 | 北京三快在线科技有限公司 | Entity identification method and device and electronic equipment |
| CN112101027A (en) * | 2020-07-24 | 2020-12-18 | 昆明理工大学 | Chinese named entity recognition method based on reading understanding |
| CN111985239A (en) * | 2020-07-31 | 2020-11-24 | 杭州远传新业科技有限公司 | Entity identification method and device, electronic equipment and storage medium |
| CN111985239B (en) * | 2020-07-31 | 2024-04-26 | 杭州远传新业科技股份有限公司 | Entity identification method, entity identification device, electronic equipment and storage medium |
| CN112287683B (en) * | 2020-08-19 | 2024-08-20 | 北京沃东天骏信息技术有限公司 | Named entity identification method and device |
| CN112287683A (en) * | 2020-08-19 | 2021-01-29 | 北京沃东天骏信息技术有限公司 | Named entity identification method and device |
| CN111708878A (en) * | 2020-08-20 | 2020-09-25 | 科大讯飞(苏州)科技有限公司 | Method, device, storage medium and equipment for extracting sports text abstract |
| CN111708878B (en) * | 2020-08-20 | 2020-11-24 | 科大讯飞(苏州)科技有限公司 | Method, device, storage medium and equipment for extracting sports text abstract |
| CN112101032A (en) * | 2020-08-31 | 2020-12-18 | 广州探迹科技有限公司 | Named entity identification and error correction method based on self-distillation |
| CN111986765B (en) * | 2020-09-03 | 2023-11-21 | 深圳平安智慧医健科技有限公司 | Electronic case entity marking method, electronic case entity marking device, electronic case entity marking computer equipment and storage medium |
| CN111986765A (en) * | 2020-09-03 | 2020-11-24 | 平安国际智慧城市科技股份有限公司 | Electronic case entity marking method, device, computer equipment and storage medium |
| WO2021151322A1 (en) * | 2020-09-07 | 2021-08-05 | 平安科技（深圳）有限公司 | Method and apparatus for entity identification based on deep learning model, device, and medium |
| CN114385809A (en) * | 2020-10-22 | 2022-04-22 | 中移(成都)信息通信科技有限公司 | Training method, device and equipment of entity text extraction model |
| CN112347773A (en) * | 2020-10-26 | 2021-02-09 | 北京诺道认知医学科技有限公司 | Medical application model training method and device based on BERT model |
| CN112765984A (en) * | 2020-12-31 | 2021-05-07 | 平安资产管理有限责任公司 | Named entity recognition method and device, computer equipment and storage medium |
| CN112906381B (en) * | 2021-02-02 | 2024-05-28 | 北京有竹居网络技术有限公司 | Dialog attribution identification method and device, readable medium and electronic equipment |
| CN112906381A (en) * | 2021-02-02 | 2021-06-04 | 北京有竹居网络技术有限公司 | Recognition method and device of conversation affiliation, readable medium and electronic equipment |
| CN112861515A (en) * | 2021-02-08 | 2021-05-28 | 上海天壤智能科技有限公司 | Interactive knowledge definition and processing method, system, device and readable medium |
| CN112801085A (en) * | 2021-02-09 | 2021-05-14 | 沈阳麟龙科技股份有限公司 | Method, device, medium and electronic equipment for recognizing characters in image |
| CN112989829B (en) * | 2021-02-10 | 2024-03-08 | 卡奥斯数字科技(上海)有限公司 | Named entity recognition method, device, equipment and storage medium |
| CN112989829A (en) * | 2021-02-10 | 2021-06-18 | 海尔数字科技（上海）有限公司 | Named entity identification method, device, equipment and storage medium |
| CN113051875B (en) * | 2021-03-22 | 2024-02-02 | 北京百度网讯科技有限公司 | Training method of information conversion model, and text information conversion method and device |
| CN113051875A (en) * | 2021-03-22 | 2021-06-29 | 北京百度网讯科技有限公司 | Training method of information conversion model, and text information conversion method and device |
| CN112966507A (en) * | 2021-03-29 | 2021-06-15 | 北京金山云网络技术有限公司 | Method, device, equipment and storage medium for constructing recognition model and identifying attack |
| CN113157883A (en) * | 2021-04-07 | 2021-07-23 | 浙江工贸职业技术学院 | Chinese opinion target boundary prediction method based on dual-model structure |
| CN113051913A (en) * | 2021-04-09 | 2021-06-29 | 中译语通科技股份有限公司 | Tibetan word segmentation information processing method, system, storage medium, terminal and application |
| CN113221565A (en) * | 2021-05-07 | 2021-08-06 | 北京百度网讯科技有限公司 | Entity recognition model training method and device, electronic equipment and storage medium |
| CN113673247A (en) * | 2021-05-13 | 2021-11-19 | 江苏曼荼罗软件股份有限公司 | Entity identification method, device, medium and electronic equipment based on deep learning |
| CN113139069A (en) * | 2021-05-14 | 2021-07-20 | 上海交通大学 | Knowledge graph construction-oriented Chinese text entity identification method and system for power failure |
| CN113204969A (en) * | 2021-05-31 | 2021-08-03 | 平安科技（深圳）有限公司 | Medical named entity recognition model generation method and device and computer equipment |
| WO2022252378A1 (en) * | 2021-05-31 | 2022-12-08 | 平安科技（深圳）有限公司 | Method and apparatus for generating medical named entity recognition model, and computer device |
| CN113408619A (en) * | 2021-06-21 | 2021-09-17 | 江苏苏云信息科技有限公司 | Language model pre-training method and device |
| CN113408619B (en) * | 2021-06-21 | 2024-02-13 | 江苏苏云信息科技有限公司 | Language model pre-training method and device |
| CN113420822A (en) * | 2021-06-30 | 2021-09-21 | 北京百度网讯科技有限公司 | Model training method and device and text prediction method and device |
| CN113849624A (en) * | 2021-10-15 | 2021-12-28 | 上海大参林医疗健康科技有限公司 | Word slot extraction device and method for multi-turn conversation |
| CN114036956A (en) * | 2021-11-18 | 2022-02-11 | 清华大学 | Tourism knowledge semantic analysis method and device |
| CN114492383A (en) * | 2021-12-20 | 2022-05-13 | 北京邮电大学 | Entity name identification method and device for digital currency transaction address |
| CN114118093A (en) * | 2022-01-27 | 2022-03-01 | 华东交通大学 | Method and system for identifying flat mark enhanced nested named entity |
| CN114564971A (en) * | 2022-02-28 | 2022-05-31 | 北京百度网讯科技有限公司 | Deep learning model training method, text data processing method and text data processing device |
| CN114648113A (en) * | 2022-03-14 | 2022-06-21 | 中移(苏州)软件技术有限公司 | Named entity recognition neural structure in reading understanding form |
| CN114706942B (en) * | 2022-03-16 | 2023-11-24 | 马上消费金融股份有限公司 | Text conversion model training method, text conversion device and electronic equipment |
| CN114706942A (en) * | 2022-03-16 | 2022-07-05 | 马上消费金融股份有限公司 | Text conversion model training method, text conversion device and electronic equipment |
| CN114707513A (en) * | 2022-03-22 | 2022-07-05 | 腾讯科技（深圳）有限公司 | Text semantic recognition method and device, electronic equipment and storage medium |
| CN116029299A (en) * | 2023-03-30 | 2023-04-28 | 北京融信数联科技有限公司 | Named entity recognition method, system and storage medium based on polysemous words |
| CN116070629A (en) * | 2023-04-06 | 2023-05-05 | 北京蜜度信息技术有限公司 | Chinese text word order checking method, system, storage medium and electronic equipment |
| CN118133830A (en) * | 2024-04-30 | 2024-06-04 | 北京壹永科技有限公司 | Named entity recognition method, named entity recognition device, named entity recognition equipment and named entity recognition computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111126068A (en) | Chinese named entity recognition method and device and electronic equipment | |
| CN112084337B (en) | Training method of text classification model, text classification method and equipment | |
| CN112131366B (en) | Method, device and storage medium for training text classification model and text classification | |
| CN113205817B (en) | Speech semantic recognition method, system, device and medium | |
| CN112528637B (en) | Text processing model training method, device, computer equipment and storage medium | |
| CN111931517B (en) | Text translation method, device, electronic equipment and storage medium | |
| CN110175336B (en) | Translation method and device and electronic equipment | |
| JP7335300B2 (en) | Knowledge pre-trained model training method, apparatus and electronic equipment | |
| CN111738016A (en) | Multi-intention recognition method and related equipment | |
| CN112131883B (en) | Language model training method, device, computer equipment and storage medium | |
| CN113158656B (en) | Ironic content recognition method, ironic content recognition device, electronic device, and storage medium | |
| CN114021582B (en) | Spoken language understanding method, device, equipment and storage medium combined with voice information | |
| CN115146068B (en) | Method, device, equipment and storage medium for extracting relation triples | |
| CN113221555A (en) | Keyword identification method, device and equipment based on multitask model | |
| CN110852066B (en) | Multi-language entity relation extraction method and system based on confrontation training mechanism | |
| CN111079418A (en) | Named body recognition method and device, electronic equipment and storage medium | |
| CN111814479B (en) | Method and device for generating enterprise abbreviations and training model thereof | |
| CN113743101A (en) | Text error correction method and device, electronic equipment and computer storage medium | |
| CN116432646A (en) | Training method of pre-training language model, entity information identification method and device | |
| CN113988061A (en) | Sensitive word detection method, device and equipment based on deep learning and storage medium | |
| CN110929532B (en) | Data processing method, device, equipment and storage medium | |
| CN112270184A (en) | Natural language processing method, device and storage medium | |
| CN115640520A (en) | Method, device and storage medium for pre-training cross-language cross-modal model | |
| CN115129862A (en) | Statement entity processing method and device, computer equipment and storage medium | |
| CN111353295A (en) | Sequence labeling method and device, storage medium and computer equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200508 |
|
| RJ01 | Rejection of invention patent application after publication |