CN113158643A - Novel text readability assessment method and system - Google Patents
Novel text readability assessment method and system Download PDFInfo
- Publication number
- CN113158643A CN113158643A CN202110459535.5A CN202110459535A CN113158643A CN 113158643 A CN113158643 A CN 113158643A CN 202110459535 A CN202110459535 A CN 202110459535A CN 113158643 A CN113158643 A CN 113158643A
- Authority
- CN
- China
- Prior art keywords
- dense
- model
- representation
- text
- dnn
- 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 24
- 238000003062 neural network model Methods 0.000 claims abstract description 25
- 238000000605 extraction Methods 0.000 claims abstract description 23
- 238000011156 evaluation Methods 0.000 claims abstract description 16
- 238000013528 artificial neural network Methods 0.000 claims abstract description 13
- 239000013598 vector Substances 0.000 claims description 20
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 4
- 102100033814 Alanine aminotransferase 2 Human genes 0.000 description 3
- 101000779415 Homo sapiens Alanine aminotransferase 2 Proteins 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/253—Grammatical analysis; Style critique
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/045—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
- G06N3/08—Learning methods
Abstract
The invention discloses a text readability assessment method and a text readability assessment system based on a two-channel neural network model, wherein the method comprises the steps of extracting semantic dense embedded expression and syntactic dense embedded expression of a text to be assessed, inputting the extracted semantic dense embedded expression and syntactic dense embedded expression into the two-channel neural network model to achieve feature extraction, combining the extracted features to form dense embedded expression of semantics and grammar, and inputting the dense embedded expression of the semantics and grammar into a deep neural network for prediction, so that the text readability assessment is achieved; by designing a two-channel neural network model, extracting semantic dense embedded representation and syntax dense embedded representation of a text to be evaluated, and evaluating from multiple angles, the efficiency and accuracy of text readability evaluation can be improved, and the defect that the evaluation is not accurate and strict enough when readability evaluation is only carried out on the aspect of word embedding level in the prior art is overcome.
Description
Technical Field
The invention relates to the technical field of text readability assessment, in particular to a novel text readability assessment method and system.
Background
The readability evaluation method aims to find language features affecting the text reading difficulty, or construct a mathematical formula or a model based on effective features to represent the language difficulty or the understanding difficulty of a text, evaluate the readability of the text, and can effectively classify the text.
However, the text readability assessment in the prior art is only expanded in terms of word embedding level, but in the text readability assessment, the structural information is crucial, one sentence with simple structure is easy to read, and another sentence with the same semantic information but complex structure is possibly more difficult to understand, so the method for expanding the readability assessment only in terms of word embedding level is not accurate and rigorous.
Disclosure of Invention
In view of this, the present invention provides a novel text readability assessment method and system, which can solve the problem of inaccurate and rigorous text readability assessment in the prior art.
The technical scheme of the invention is realized as follows:
a novel text readability assessment method based on a two-channel neural network model comprises the following steps:
step S1, extracting semantic dense embedded representation and syntax dense embedded representation of the text to be evaluated;
step S2, inputting the extracted semantic dense embedded expression and syntax dense embedded expression into a two-channel neural network model to realize the extraction of features;
step S3, combining the extracted features to form dense embedded expression of semantics and grammar;
and step S4, inputting the dense embedded expression of the semantics and the grammar into a deep neural network for prediction, thereby realizing the evaluation of the readability of the text.
As a further alternative of the novel text readability assessment method, the step S1 of extracting the semantically dense embedded representation of the text to be assessed is generated by using a pre-trained semantic model.
As a further alternative of the novel text readability assessment method, the step of extracting the syntactic dense embedded representation of the text to be assessed in step S1 includes the steps of:
extracting language features of a text to be evaluated and constructing vector space representation;
constructing a feature graph according to the relation between each language feature;
applying a representation learning algorithm to the feature map to generate a syntactic latent space;
the language feature vectors are projected onto the syntactic potential space using matrix-vector multiplication, forming a syntactic dense embedded representation.
As a further alternative of the novel text readability assessment method, the two-channel neural network model comprises a DNN two-channel model and a DNN-LSTM two-channel model.
As a further alternative of the novel text readability assessment method, the step S2 of extracting features of the syntax dense embedding representation adopts a DNN two-channel model, and mainly includes the following steps:
inputting the syntactic dense embedded representation into a two-layer network of the DNN two-channel model;
and combining two layers of output contents of the DNN dual-channel model with four layers of neural network output contents of the pre-trained semantic model embedded matrix according to the average embedded representation of the pre-trained semantic model, thereby realizing the feature extraction of syntax dense embedded representation.
As a further alternative of the novel text readability assessment method, the feature extraction of the dense semantic embedded representation in step S2 is performed by using a DNN-LSTM two-channel model, which mainly includes the following steps:
embedding the pre-training semantic model into a BilSTM network to form a DNN-LSTM dual-channel model;
inputting the semantically dense embedded representation into the DNN-LSTM two-channel model, and outputting the features of the semantically dense embedded representation by the DNN-LSTM two-channel model.
A novel text readability assessment system, said system comprising:
the first extraction module is used for extracting semantic dense embedded representation and syntactic dense embedded representation of the text to be evaluated;
the two-channel neural network model is used for carrying out feature extraction on the semantic dense embedded representation and the syntactic dense embedded representation;
the combination module is used for combining the extracted features to form dense embedded expression of semantics and grammar;
and the deep neural network model is used for evaluating the readability of the text according to the dense embedded expression of the semantics and the grammar.
As a further alternative of the novel text readability assessment system, the first extraction module comprises:
the storage module is used for storing the pre-training semantic model;
the first construction module is used for extracting the language features of the text to be evaluated and constructing vector space representation;
a second construction module for constructing a feature map based on the relationship between each language feature
The first generation module is used for applying a representation learning algorithm to the feature diagram to generate a syntax potential space;
a second generation module to project the language feature vectors onto a syntactic latent space using matrix-vector multiplication, thereby generating a syntactic dense embedded representation.
As a further alternative to the new text readability assessment system, the two-channel neural network model includes a DNN two-channel model and a DNN-LSTM two-channel model.
As a further alternative to the new text readability assessment system, the DNN two-channel model comprises:
an input module for inputting the syntactic dense embedded representation into a two-layer network of the DNN two-channel model;
and the feature extraction module is used for combining two layers of output contents of the DNN dual-channel model with four layers of neural network output contents of the pre-training semantic model embedded matrix according to the average embedded representation of the pre-training semantic model so as to extract features of syntax dense embedded representation.
The invention has the beneficial effects that: by designing a two-channel neural network model, extracting semantic dense embedded representation and syntax dense embedded representation of a text to be evaluated, and evaluating from multiple angles, the efficiency and accuracy of text readability evaluation can be improved, and the defect that the evaluation is not accurate and strict enough when readability evaluation is only carried out on the aspect of word embedding level in the prior art is overcome.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of a novel text readability assessment method according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all 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.
Referring to fig. 1, a novel text readability assessment method based on a two-channel neural network model includes the following steps:
step S1, extracting semantic dense embedded representation and syntax dense embedded representation of the text to be evaluated;
step S2, inputting the extracted semantic dense embedded expression and syntax dense embedded expression into a two-channel neural network model to realize the extraction of features;
step S3, combining the extracted features to form dense embedded expression of semantics and grammar;
and step S4, inputting the dense embedded expression of the semantics and the grammar into a deep neural network for prediction, thereby realizing the evaluation of the readability of the text.
In the embodiment, by designing a dual-channel neural network model, extracting semantic dense embedded representation and syntax dense embedded representation of the text to be evaluated, and evaluating from multiple angles, the efficiency and accuracy of text readability evaluation can be improved, and the defect that the evaluation is not accurate and strict enough in the prior art that readability evaluation is only performed on the aspect of word embedding level is overcome.
Preferably, the step S1 of extracting the semantically dense embedded representation of the text to be evaluated is generated by using a pre-trained semantic model.
In this embodiment, the pre-training semantic model includes, but is not limited to, a Robert model, an XLnet model, a BERT model, a GPT2 model, and a GPT3 model, which are not limited herein.
Preferably, the extracting of the syntax dense embedded representation of the text to be evaluated in the step S1 includes the following steps:
extracting language features of a text to be evaluated and constructing vector space representation;
constructing a feature graph according to the relation between each language feature;
applying a representation learning algorithm to the feature map to generate a syntactic latent space;
the language feature vectors are projected onto the syntactic potential space using matrix-vector multiplication, forming a syntactic dense embedded representation.
In the present embodiment, the representation learning algorithm includes, but is not limited to, a TransE algorithm and an analog algorithm, and is not limited to this.
Preferably, the two-channel neural network model includes a DNN two-channel model and a DNN-LSTM two-channel model.
Preferably, the step S2 of extracting features of the syntactic dense embedded representation is to use a DNN two-channel model, which mainly includes the following steps:
inputting the syntactic dense embedded representation into a two-layer network of the DNN two-channel model;
and combining two layers of output contents of the DNN dual-channel model with four layers of neural network output contents of the pre-trained semantic model embedded matrix according to the average embedded representation of the pre-trained semantic model, thereby realizing the feature extraction of syntax dense embedded representation.
Preferably, the feature extraction of the dense semantic embedded representation in step S2 is performed by using a DNN-LSTM two-channel model, which mainly includes the following steps:
embedding the pre-training semantic model into a BilSTM network to form a DNN-LSTM dual-channel model;
inputting the semantically dense embedded representation into the DNN-LSTM two-channel model, and outputting the features of the semantically dense embedded representation by the DNN-LSTM two-channel model.
Example 1:
representing document features using two dense embedded representation methods, namely a semantically dense embedded representation and a syntactically dense embedded representation, the semantically dense embedded representation being an embedded representation generated using a pre-trained semantic model that can learn dense representations of text by semantically neighboring words, sentences or contexts, the pre-trained semantic model including, but not limited to, Robert model, XLnet model, BERT model, GPT2 model and GPT3 model, the syntactically dense embedded representation being learned from linguistic features that explicitly capture structural features of text, in particular by first extracting linguistic features and constructing a vector space representation for each document, then building a feature map using correlations between features and learning linguistic feature embedding using the TransE algorithm and the anolog y algorithm, then projecting the document into a linguistic potential space using matrix vector multiplication, obtaining a document projection in a syntactically potential space, and finally, performing further feature extraction on the dense embedded expression of the semantics and the grammar through a neural network so as to realize readability prediction.
It should be noted that, the two-channel neural network models proposed by the present invention have two types, which are slightly different, one is a DNN two-channel model with enhanced document level representation, and the other is a DNN-LSTM two-channel model with enhanced document level representation; the DNN dual-channel model firstly inputs syntax dense embedded representation into a two-layer network, simultaneously learns BERT average embedded representation of a document, and then combines two layers of output content with four layers of neural network output content of a BERT embedded matrix to form semantic and syntax dense embedded representation as input to the other two-layer network, wherein MLP and SoftMax layers are used for readability classification; the DNN-LSTM two-channel model, which is slightly different, embeds BERT into the BiLSTM network and then appends its output to the DNN output of the syntactic dense embedded representation to form a dense embedded representation of semantics and grammar, which performs further feature extraction and document readability level prediction through a neural network to generate output content juxtaposed to the DNN output and document projection.
A novel text readability assessment system, said system comprising:
the first extraction module is used for extracting semantic dense embedded representation and syntactic dense embedded representation of the text to be evaluated;
the two-channel neural network model is used for carrying out feature extraction on the semantic dense embedded representation and the syntactic dense embedded representation;
the combination module is used for combining the extracted features to form dense embedded expression of semantics and grammar;
and the deep neural network model is used for evaluating the readability of the text according to the dense embedded expression of the semantics and the grammar.
In the embodiment, by designing a dual-channel neural network model, extracting semantic dense embedded representation and syntax dense embedded representation of the text to be evaluated, and evaluating from multiple angles, the efficiency and accuracy of text readability evaluation can be improved, and the defect that the evaluation is not accurate and strict enough in the prior art that readability evaluation is only performed on the aspect of word embedding level is overcome.
Preferably, the first extraction module includes:
the storage module is used for storing the pre-training semantic model;
the first construction module is used for extracting the language features of the text to be evaluated and constructing vector space representation;
a second construction module for constructing a feature map based on the relationship between each language feature
The first generation module is used for applying a representation learning algorithm to the feature diagram to generate a syntax potential space;
a second generation module to project the language feature vectors onto a syntactic latent space using matrix-vector multiplication, thereby generating a syntactic dense embedded representation.
In the present embodiment, the pre-training semantic model includes, but is not limited to, Robert model, XLnet model, BERT model, GPT2 model, and GPT3 model, and the representation learning algorithm includes, but is not limited to, TransE algorithm and analoy algorithm, which are not limited herein.
Preferably, the two-channel neural network model includes a DNN two-channel model and a DNN-LSTM two-channel model.
Preferably, the DNN two-channel model includes:
an input module for inputting the syntactic dense embedded representation into a two-layer network of the DNN two-channel model;
and the feature extraction module is used for combining two layers of output contents of the DNN dual-channel model with four layers of neural network output contents of the pre-training semantic model embedded matrix according to the average embedded representation of the pre-training semantic model so as to extract features of syntax dense embedded representation.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. A novel text readability assessment method is characterized in that the method is based on a dual-channel neural network model and comprises the following steps:
step S1, extracting semantic dense embedded representation and syntax dense embedded representation of the text to be evaluated;
step S2, inputting the extracted semantic dense embedded expression and syntax dense embedded expression into a two-channel neural network model to realize the extraction of features;
step S3, combining the extracted features to form dense embedded expression of semantics and grammar;
and step S4, inputting the dense embedded expression of the semantics and the grammar into a deep neural network for prediction, thereby realizing the evaluation of the readability of the text.
2. The method as claimed in claim 1, wherein the step S1 of extracting the semantically dense embedded representation of the text to be evaluated is generated by using a pre-trained semantic model.
3. The method of claim 2, wherein the step S1 of extracting the syntactically dense embedded representation of the text to be evaluated comprises the steps of:
extracting language features of a text to be evaluated and constructing vector space representation;
constructing a feature graph according to the relation between each language feature;
applying a representation learning algorithm to the feature map to generate a syntactic latent space;
the language feature vectors are projected onto the syntactic potential space using matrix-vector multiplication, forming a syntactic dense embedded representation.
4. The method of claim 3, wherein the two-channel neural network model comprises a DNN two-channel model and a DNN-LSTM two-channel model.
5. The method for evaluating readability of new texts as claimed in claim 4, wherein said step S2 of extracting features of said expression of dense embedding of sentence method is implemented by using a DNN two-channel model, which mainly includes the following steps:
inputting the syntactic dense embedded representation into a two-layer network of the DNN two-channel model;
and combining two layers of output contents of the DNN dual-channel model with four layers of neural network output contents of the pre-trained semantic model embedded matrix according to the average embedded representation of the pre-trained semantic model, thereby realizing the feature extraction of syntax dense embedded representation.
6. The method as claimed in claim 5, wherein the step S2 of extracting features of the dense semantic embedded representation is implemented by using a DNN-LSTM two-channel model, and comprises the following steps:
embedding the pre-training semantic model into a BilSTM network to form a DNN-LSTM dual-channel model;
inputting the semantically dense embedded representation into the DNN-LSTM two-channel model, and outputting the features of the semantically dense embedded representation by the DNN-LSTM two-channel model.
7. A novel text readability assessment system, said system comprising:
the first extraction module is used for extracting semantic dense embedded representation and syntactic dense embedded representation of the text to be evaluated;
the two-channel neural network model is used for carrying out feature extraction on the semantic dense embedded representation and the syntactic dense embedded representation;
the combination module is used for combining the extracted features to form dense embedded expression of semantics and grammar;
and the deep neural network model is used for evaluating the readability of the text according to the dense embedded expression of the semantics and the grammar.
8. The system of claim 7, wherein the first extraction module comprises:
the storage module is used for storing the pre-training semantic model;
the first construction module is used for extracting the language features of the text to be evaluated and constructing vector space representation;
a second construction module for constructing a feature map based on the relationship between each language feature
The first generation module is used for applying a representation learning algorithm to the feature diagram to generate a syntax potential space;
a second generation module to project the language feature vectors onto a syntactic latent space using matrix-vector multiplication, thereby generating a syntactic dense embedded representation.
9. The system of claim 8, wherein the two-channel neural network model comprises a DNN two-channel model and a DNN-LSTM two-channel model.
10. The system of claim 9, wherein the DNN two-channel model comprises:
an input module for inputting the syntactic dense embedded representation into a two-layer network of the DNN two-channel model;
and the feature extraction module is used for combining two layers of output contents of the DNN dual-channel model with four layers of neural network output contents of the pre-training semantic model embedded matrix according to the average embedded representation of the pre-training semantic model so as to extract features of syntax dense embedded representation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110459535.5A CN113158643A (en) | 2021-04-27 | 2021-04-27 | Novel text readability assessment method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110459535.5A CN113158643A (en) | 2021-04-27 | 2021-04-27 | Novel text readability assessment method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113158643A true CN113158643A (en) | 2021-07-23 |
Family
ID=76871422
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110459535.5A Pending CN113158643A (en) | 2021-04-27 | 2021-04-27 | Novel text readability assessment method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113158643A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117313704A (en) * | 2023-11-28 | 2023-12-29 | 江西师范大学 | Mixed readability evaluation method and system based on public and private feature decomposition |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6374217B1 (en) * | 1999-03-12 | 2002-04-16 | Apple Computer, Inc. | Fast update implementation for efficient latent semantic language modeling |
| US20020087316A1 (en) * | 2000-12-29 | 2002-07-04 | Lee Victor Wai Leung | Computer-implemented grammar-based speech understanding method and system |
| US20040215449A1 (en) * | 2002-06-28 | 2004-10-28 | Philippe Roy | Multi-phoneme streamer and knowledge representation speech recognition system and method |
| CN107291822A (en) * | 2017-05-24 | 2017-10-24 | 北京邮电大学 | The problem of based on deep learning disaggregated model training method, sorting technique and device |
| CN109684626A (en) * | 2018-11-16 | 2019-04-26 | 深思考人工智能机器人科技（北京）有限公司 | Method for recognizing semantics, model, storage medium and device |
| CN110472616A (en) * | 2019-08-22 | 2019-11-19 | 腾讯科技（深圳）有限公司 | Image-recognizing method, device, computer equipment and storage medium |
| CN111008526A (en) * | 2019-12-06 | 2020-04-14 | 安徽理工大学 | Named entity identification method based on dual-channel neural network |
-
2021
- 2021-04-27 CN CN202110459535.5A patent/CN113158643A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6374217B1 (en) * | 1999-03-12 | 2002-04-16 | Apple Computer, Inc. | Fast update implementation for efficient latent semantic language modeling |
| US20020087316A1 (en) * | 2000-12-29 | 2002-07-04 | Lee Victor Wai Leung | Computer-implemented grammar-based speech understanding method and system |
| US20040215449A1 (en) * | 2002-06-28 | 2004-10-28 | Philippe Roy | Multi-phoneme streamer and knowledge representation speech recognition system and method |
| CN107291822A (en) * | 2017-05-24 | 2017-10-24 | 北京邮电大学 | The problem of based on deep learning disaggregated model training method, sorting technique and device |
| CN109684626A (en) * | 2018-11-16 | 2019-04-26 | 深思考人工智能机器人科技（北京）有限公司 | Method for recognizing semantics, model, storage medium and device |
| CN110472616A (en) * | 2019-08-22 | 2019-11-19 | 腾讯科技（深圳）有限公司 | Image-recognizing method, device, computer equipment and storage medium |
| CN111008526A (en) * | 2019-12-06 | 2020-04-14 | 安徽理工大学 | Named entity identification method based on dual-channel neural network |
Non-Patent Citations (4)
| Title |
|---|
| YUAN LUO ETC: "Bridging Semantics and Syntax with Graph Algorithms - State-of-the-Art of Extracting Biomedical Relations", BRIEFINGS IN BIOINFORMATICS, vol. 18, no. 1, pages 160 - 178 * |
| 丘心颖 等: "融合 Self-Attention 机制和 n-gram 卷积核的印尼语复合名词自动识别方法研究", 湖南工业大学学报, vol. 34, no. 3, pages 1 - 9 * |
| 邱宁佳 等: "改进 CNN的多通道语义合成情感分类模型研究", 计算机工程与应用, vol. 55, no. 23, pages 136 - 141 * |
| 邱宁佳 等: "融合语法规则的双通道中文情感模型分析", 计算机应用, vol. 41, no. 2, pages 318 - 323 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117313704A (en) * | 2023-11-28 | 2023-12-29 | 江西师范大学 | Mixed readability evaluation method and system based on public and private feature decomposition |
| CN117313704B (en) * | 2023-11-28 | 2024-02-23 | 江西师范大学 | Mixed readability evaluation method and system based on public and private feature decomposition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rocktäschel et al. | Reasoning about entailment with neural attention | |
| JP7072585B2 (en) | Natural language processing with context-specific word vectors | |
| US20200242146A1 (en) | Artificial intelligence system for generating conjectures and comprehending text, audio, and visual data using natural language understanding | |
| CN114064918A (en) | Multi-modal event knowledge graph construction method | |
| CN113378547B (en) | GCN-based Chinese complex sentence implicit relation analysis method and device | |
| CN110245349B (en) | Syntax dependence analysis method and apparatus, and electronic device | |
| CN114519356B (en) | Target word detection method and device, electronic equipment and storage medium | |
| CN113392197A (en) | Question-answer reasoning method and device, storage medium and electronic equipment | |
| CN114996423A (en) | Water conservancy knowledge graph complex question-answering method based on subproblem pruning | |
| CN109815497B (en) | Character attribute extraction method based on syntactic dependency | |
| Mezghanni et al. | Deriving ontological semantic relations between Arabic compound nouns concepts | |
| CN113158643A (en) | Novel text readability assessment method and system | |
| CN116661805B (en) | Code representation generation method and device, storage medium and electronic equipment | |
| CN112883713A (en) | Evaluation object extraction method and device based on convolutional neural network | |
| CN110377753B (en) | Relation extraction method and device based on relation trigger word and GRU model | |
| Loginova et al. | Structural information in mathematical formulas for exercise difficulty prediction: a comparison of NLP representations | |
| CN116702746A (en) | Cross-platform multi-theme irony and cause identification method, device, equipment and medium | |
| CN113449517B (en) | Entity relationship extraction method based on BERT gated multi-window attention network model | |
| CN111813927A (en) | Sentence similarity calculation method based on topic model and LSTM | |
| JP6558856B2 (en) | Morphological analyzer, model learning device, and program | |
| CN116468030A (en) | End-to-end face-level emotion analysis method based on multitasking neural network | |
| Zheng | A Novel Computer-Aided Emotion Recognition of Text Method Based on WordEmbedding and Bi-LSTM | |
| CN112349294A (en) | Voice processing method and device, computer readable medium and electronic equipment | |
| Rajaraman et al. | Deepreply-an automatic email reply system with unsupervised cloze translation and deep learning | |
| CN116227484B (en) | Model training method, apparatus, device, storage medium and computer program product |
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 |