A Method for Logical Coherence Analysis of English Text

Abstract

This invention provides a method for analyzing the logical coherence of English text. This method comprises a module for generating a sequentially connected logical relationship tree of English text, a module for transforming an English text logical relationship diagram, and a module for analyzing the logical coherence of English text. After processing an English text using this method, the logical coherence analysis results can be obtained. This invention solves the problem that existing English text coherence analysis methods can only analyze the lexical coherence of English text, but cannot analyze its logical coherence.

Description

Technical Field

[0001] This invention relates to natural language processing technology and is a method for analyzing the logical coherence of English text. The method of this invention is only suitable for analyzing the logical coherence quality of English text and is not suitable for other logical coherence quality analyses. Background Technology

[0002] Existing methods for analyzing the coherence of English text extract coherence information from words and entities and then use machine learning and deep learning to determine the degree of coherence. However, these methods can only analyze lexical coherence, not logical coherence. To address this issue, this invention proposes a method for analyzing the logical coherence of English text. This method analyzes the quality of logical coherence of English text by starting from its logical relationships. Summary of the Invention

[0003] The present invention provides a method for analyzing the logical coherence of English text, comprising: an English text logical relation tree generation module, an English text logical relation diagram conversion module, and an English text logical coherence analysis module. The overall processing flowchart is shown below. Figure 1 As shown.

[0004] The processing flow of the English text logical relation tree generation module of the present invention is as follows: First, read the English text to be analyzed and perform sentence segmentation and word segmentation on the English text to be analyzed; Second, perform logical relation parsing on the English text after sentence and word segmentation and construct an English text logical relation tree; Third, output the constructed English text logical relation tree.

[0005] The processing flow of the English text logical relationship graph conversion module of the present invention is as follows: First, read the English text logical relationship tree; Second, traverse the English text logical relationship tree, and construct the edges and nodes of the English text logical relationship graph according to the logical relationships in the English text logical relationship tree. The edges are logical relationships, and the nodes connected by the edges are the core clauses of the first and second subtrees of the logical relationship. If there is no core clause, then the leftmost non-core clause of the first and second subtrees is selected as the core clause; Third, output the constructed English text logical relationship graph.

[0006] The processing flow of the English text logical coherence analysis module of this invention is as follows: First, read the English text logical relationship graph; Second, based on the English text logical relationship graph, construct the adjacency grid and feature grid of the English text logical relationship graph using the English text logical relationship graph adjacency grid calculation formula and the English text logical relationship graph feature grid calculation formula, respectively. The English text logical relationship graph adjacency grid represents the connection relationship between nodes in the English text logical relationship graph, and the English text logical relationship graph feature grid represents the feature information of nodes in the English text logical relationship graph; Third, encode the English text logical relationship graph adjacency grid and the English text logical relationship graph feature grid, and calculate the English text graph logical coherence vector using the English text logical relationship graph coherence vector calculation formula; Fourth, based on the English text graph logical coherence vector, calculate the English text logical coherence score using the English text logical coherence score calculation formula, and output the English text logical coherence score and comments.

[0007] The concept and calculation formula of this invention

[0008] (1) Formula for calculating the adjacency grid of the logical relationship graph in English text

[0009]

[0010] In formula (1), i and j represent the i-th node and the j-th node in the English text logical relationship graph, respectively. The positions of the i-th node and the j-th node in the adjacency grid of the English text logical relationship graph are represented by i and j. When there is an edge between node i and node j, the corresponding value in the adjacency grid of the English text logical relationship graph is 1, otherwise it is 0.

[0011] (2) Formula for calculating the feature grid of the logical relationship diagram in English text

[0012]

[0013] In formula (2), n represents the total number of nodes in the logical relationship graph of the English text, w represents the word order information of the node, and m represents the total number of words in the node.

[0014] (3) Formula for calculating the coherence vector of the logical relationship diagram in English text

[0015]

[0016] In formula (3), i represents the i-th row corresponding to the i-th node in the adjacency grid and feature grid of the English text logical relationship graph, and n represents the total number of nodes in the English text logical relationship graph. The adjacency grid of the English text logical relationship graph is obtained by formula (1), and the feature grid of the English text logical relationship graph is obtained by formula (2).

[0017] (4) Formula for calculating the logical coherence score of English text

[0018]

[0019] In formula (4), j represents the j-th number in the coherent vector of the English text logic diagram, m represents the total dimension of the coherent vector of the English text logic diagram, and e represents the base of the natural logarithm function. The coherent vector of the English text logic diagram is calculated by formula (3).

[0020] Specific processing steps of the correction method of the present invention

[0021] like Figure 1 As shown in the diagram, the overall processing flow is as follows:

[0022] Starting from page 101;

[0023] P102 executes the English text logical relation tree generation module, the specific steps are as follows: Figure 2 As shown;

[0024] Page 103 executes the English text logic diagram conversion module. Specific steps are as follows: Figure 3 As shown;

[0025] Page 103 executes the English text logical coherence analysis module; the specific steps are as follows: Figure 4 As shown;

[0026] Page 105 ends.

[0027] like Figure 2 As shown, the processing flow of the English text logical relation tree generation module is as follows:

[0028] Starting from page 201;

[0029] P202 reads the English text to be analyzed;

[0030] P203 Sentence segmentation of the English text to be analyzed;

[0031] P204 performs word segmentation on the English text to be analyzed;

[0032] P205 analyzes the logical relationships of the English text after sentence and word segmentation, and constructs a logical relationship tree for the English text.

[0033] P206 Output the constructed English text logical relation tree;

[0034] End of page 207.

[0035] like Figure 3 As shown, the processing flow of the English text logical relationship diagram conversion module is as follows:

[0036] Starting from P301;

[0037] P302 reads the logical relationship tree of English text;

[0038] P303 Initializes the logical relationship queue;

[0039] P304 inputs the root node of the logical relation tree of the English text into the logical relation queue;

[0040] P305 performs a dequeue operation on the head node of the logical relation queue and sets it as the current node;

[0041] P306 checks if the root node of the first subtree of the current node is empty; otherwise, proceed to operation P307; otherwise, proceed to operation P308.

[0042] P307 inputs the root node of the first subtree of the current node into the logical relation queue;

[0043] P308 checks if the root node of the second subtree of the current node is empty; otherwise, proceed to operation P309; otherwise, proceed to operation P310.

[0044] P309 inputs the root node of the second subtree of the current node into the logical relation queue;

[0045] P310 checks whether the current node is a leaf node; otherwise, proceed to operation P311; otherwise, proceed to operation P313.

[0046] P311 sets the core clause of the first subtree of the current node as a node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause.

[0047] P312 sets the core clause of the second subtree of the current node as the other node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause, and then the operation proceeds to P313.

[0048] P313 checks if the logical relation queue is empty; otherwise, proceed to P305; otherwise, proceed to P314.

[0049] Output the logical relationship diagram of the English text on page 314;

[0050] End of P315.

[0051] like Figure 4 As shown, the processing flow of the English text logical coherence analysis module is as follows:

[0052] Starting from P401;

[0053] P402 Read the logical relationship diagram of English text;

[0054] P403 Calculate the adjacency grid of the English text logical relationship graph according to the adjacency grid formula (1);

[0055] Page 404 uses feature vectors to represent the nodes of the logical relationship graph of English text.

[0056] P405 Calculate the feature grid of the English text logic graph according to the feature grid formula (2);

[0057] P406 Calculate the coherence vector of the English text logic diagram according to the formula (3) for the coherence vector of the English text logic diagram;

[0058] P407 Calculate the logical coherence score of English text according to the English text logical coherence score formula (4);

[0059] P408 generates comments on the logical coherence of English text;

[0060] End of page 409.

[0061] like Figure 5 As shown, the logical relationships of the English text obtained from the text to be analyzed are as follows:

[0062] In the graph, nodes represent clauses, and edges between nodes represent the relationships between two clauses. Specifically:

[0063] Generalization relationships: There are generalization relationships between clause 1 and clause 2, clause 6 and clause 8, clause 8 and clause 11, and clause 13 and clause 18.

[0064] The following clauses exhibit explanatory relationships: Clause 2 and Clause 3, Clause 6 and Clause 7, Clause 11 and Clause 12, Clause 13 and Clause 14, Clause 14 and Clause 15, Clause 13 and Clause 17, Clause 6 and Clause 20, and Clause 20 and Clause 21.

[0065] Background relationships: There are background relationships between clauses 4 and 5, and between clauses 19 and 20;

[0066] Union: Clause 2 and Clause 5, and Clause 8 and Clause 13 are in a union relationship;

[0067] Problem-solving relationship: Clause 1 and Clause 6 have a problem-solving relationship;

[0068] Method relationship: Clause 8 and Clause 9 have a method relationship;

[0069] List relationship: Clause 9 and Clause 10 are in a list relationship;

[0070] Purpose relationship: There is a purpose relationship between clause 16 and clause 17. Attached Figure Description

[0071] Figure 1 This is the overall processing flowchart of the present invention;

[0072] Figure 2 This is a flowchart of the English text logical relation tree generation module of the present invention;

[0073] Figure 3 This is a flowchart of the English text logic diagram conversion module of the present invention;

[0074] Figure 4 This is a flowchart of the English text logical coherence analysis module of the present invention;

[0075] Figure 5 This is the English text logic diagram of the present invention. Detailed Implementation

[0076] The present invention provides a method for analyzing the logical coherence of English text, comprising the following steps:

[0077] First step: Execute the "English Text Logical Relationship Tree Generation Module"

[0078] (1) The English text to be analyzed is as follows:

[0079] It is very important for college students to get to know the worldoutside the campus.For one thing,they can learn much that cannot be got frombooks from the world outside the campus.For another,as college students,getting to know the society will enable them to adapt to the society muchbetter after their graduation.There are several ways to get to know theworld,such as mediums and social services.First,college students can get toknow the world by reading newspaper,listening to radio and watchingtelevision.These mediums display a complex society before us.This is anindirect way.Second,we should go out of the campus and go into the society,take part in some practical social activity.For instance,we can go to theremote countryside to help the villagers develop education.Such practicalactivity and experience will make us have a further understanding about theworld.As for me,I will get to know the world outside the campus from mediums,and go into the society in my holidays.

[0080] (2) Logical relationship analysis was performed on the English text after sentence and word segmentation, and the resulting logical relationship tree of the English text is as follows:

[0081]

[0082]

[0083] Step 2: Execute the "English Text Logical Relationship Diagram Conversion Module" to output the converted English text logical relationships, as shown below. Figure 5 As shown.

[0084] Step 3: Execute the "English Text Logical Coherence Analysis Module"

[0085] (1) The adjacency grid of the logical relationship diagram of English text is as follows:

[0086] [[0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0087] [1,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0088] [0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0089] [0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0090] [0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0091] [1,0,0,0,0,0,1,1,0,0,0,0,0,0,0,0,0,0,1,0],

[0092] [0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0093] [0,0,0,0,0,1,0,0,1,0,1,0,1,0,1,0,0,0,0,0,0,0],

[0094] [0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0],

[0095] [0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0],

[0096] [0,0,0,0,0,0,0,1,0,0,0,1,0,0,0,0,0,0,0,0,0],

[0097] [0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0],

[0098] [0,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,1,1,0,0,0],

[0099] [0,0,0,0,0,0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,0],

[0100] [0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0],

[0101] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0],

[0102] [0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0,0,0,0,0],

[0103] [0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0],

[0104] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0],

[0105] [0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,1,0,1],

[0106] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0]]

[0107] (2) The words in the English text to be analyzed are represented by vectors, and the resulting word vectors are as follows:

[0108] The first clause

[0109] [0.03328400105237961,-0.04075400158762932,...,-0.19683000445365906],[-0.17489999532699585,0.2295600026845932,...,0.36812999844551086],

[0110] [-0.19022999703884125,0.11810000240802765,...,0.10862000286579132],[0.237419992685318,0.10226999968290329,...,0.16607999801635742],

[0111] [-0.24131999909877777,0.12063000351190567,...,0.1550700068473816],

[0112] [-0.501479983329773,0.4590800106525421,...,0.22214999794960022],

[0113] [-0.4538100063800812,-0.007326600141823292,...,-0.1926800012588501],[-0.25756001472473145,-0.05713199824094772,...,-0.07062099874019623],[-0.14124000072479248,-0.1183599978685379,...,0.11568000167608261],[-0.25756001472473145,-0.05713199824094772,...,-0.07062099874019623],[-0.21053999662399292,0.13819999992847443,...,0.5273299813270569],

[0114] [0.046560000628232956,0.21318000555038452,...,0.05391300097107887],

[0115] [-0.2583099901676178,0.43643999099731445,...,-0.1410199999809265],

[0116] [-0.07607100158929825,0.34909000992774963,...,0.261029988527298],

[0117] [0.046560000628232956,0.21318000555038452,...,0.05391300097107887],[-0.16909000277519226,-0.37292999029159546,...,0.005296700168401003]

[0118] The second clause

[0119] [-0.24131999909877777,0.12063000351190567,...,0.1550700068473816],

[0120] [-0.3675599992275238,0.39500001072883606,...,0.0314910002052784],

[0121] [0.2646400034427643, -0.10502000153064728, ..., 0.4142700135707855],

[0122] [-0.26625001430511475,0.47964000701904297,...,-0.004142399877309799],[-0.5110200047492981,0.6175199747085571,...,-0.18386000394821167],

[0123] [-0.33156999945640564,-0.08879599720239639,...,0.001311299973167479],[-0.20117999613285065,-0.13655999302864075,...,-0.1729699969291687]

[0124] ...

[0125] 21st clause

[0126] [0.03846599906682968,-0.039792001247406006,...,0.059702999889850616],[0.007367799989879131,0.06253200024366379,...,0.2936199903488159],

[0127] [-0.18233999609947205,-0.1920900046825409,...,0.176829993724823],

[0128] [0.046560000628232956,0.21318000555038452,...,0.05391300097107887],

[0129] [-0.10101000219583511,-0.3196200132369995,...,-0.15689000487327576],

[0130] [-0.44398999214172363,0.12816999852657318,...,-0.06255000084638596],

[0131] [-0.22745999693870544,-0.13658000528812408,...,0.5534600019454956],

[0132] [-0.009709199890494347,-0.31836000084877014,...,-0.4344399869441986]

[0133] (3) The feature grid of the logical relationship diagram of English text is as follows:

[0134] [[-2.568287007510662,1.726095444522798,...,1.0520546776242554],

[0135] [-1.6542600095272064,1.2824140042066574,...,0.24116991984192282],

[0136] [-1.7835206910967827,1.3384200185537338,...,-0.29846930457279086],

[0137] [-1.5155160017311573,1.0635534031316638,...,0.49550001323223114],

[0138] [-2.885171040892601,0.5408970136195421,...,-1.3463520868681371],

[0139] [-1.2205199748277664,0.5272980108857155,...,-0.3073039874434471],

[0140] [-2.831090033054352,0.37104197312146425,...,0.11796198971569538],

[0141] [-2.443009991198778,2.1822913689538836,...,0.4753919802606106],

[0142] [-0.6046049892902374,0.3198600113391876,...,-1.0426799952983856],

[0143] [0.24025099352002144,0.7467179894447327,...,0.31605999544262886],

[0144] [-1.4510132793802768,0.5167669774964452,...,-0.8604817083105445],

[0145] [-0.7387700080871582,0.7229420095682144,...,0.2152549959719181],

[0146] [-0.2675502044148743,0.7188690304756165,...,0.4160376996733248],

[0147] [-0.1909561986103654,-0.2757900133728981,...,0.42717598006129265],

[0148] [-1.1180119942873716,0.09529000520706177,...,-0.9624240007251501],

[0149] [-2.0874872291460633,1.272486001253128,...,0.29737259610556066],

[0150] [-1.61260100081563,0.2442139983177185,...,0.04682500660419464],

[0151] [-1.7310298702213913,1.0048769786953926,...,-0.19307767506688833],

[0152] [-0.302187598310411,0.10571999102830887,...,0.7520700246095657],

[0153] [-2.2252810150384903,0.7706560343503952,...,1.2530716476030648],

[0154] [-0.8721153875812888,-0.6025600209832191,...,0.4836459942162037]]

[0155] (4) The scores and comments for the logical coherence analysis of the English text are as follows:

[0156] Based on formulas (3) and (4), the logical coherence of English texts is analyzed, and the specific analysis results are shown below:

[0157] Logical coherence score: 0.673

[0158] Comments on logical coherence: The English text to be analyzed has a generally natural transition, relatively clear logic, and good coherence.

Claims

1. A method for analyzing the logical coherence of English texts, characterized by: It consists of a module for generating a logical relationship tree of sequentially connected English text, a module for transforming a logical relationship diagram of English text, and a module for analyzing the logical coherence of English text. Its analysis method includes the following processing flow: 1) The processing flow of the English text logical relation tree generation module is as follows: read the English text to be analyzed, perform sentence segmentation and word segmentation on the English text to be analyzed; perform logical relation parsing on the English text after sentence and word segmentation, and construct the English text logical relation tree; Output the constructed logical relation tree of the English text; 2) The processing flow of the English text logical relationship diagram conversion module is: read the English text logical relationship tree; Traverse the logical relation tree of the English text. Based on the logical relations in the logical relation tree, construct the edges and nodes of the logical relation graph of the English text. The edges are logical relations, and the nodes connected by the edges are the core clauses of the first and second subtrees of the logical relations. If there is no core clause, then select the leftmost non-core clause of the first and second subtrees as the core clause; output the constructed logical relation graph of the English text. 3) The processing flow of the English text logical coherence analysis module is as follows: Read the English text logical relationship graph; based on the English text logical relationship graph, construct the adjacency grid and feature grid of the English text logical relationship graph using the adjacency grid calculation formula and the feature grid calculation formula, respectively. The adjacency grid represents the connection relationship between nodes in the English text logical relationship graph, and the feature grid represents the feature information of the nodes in the English text logical relationship graph; encode the adjacency grid and feature grid of the English text logical relationship graph, and calculate the graph logical coherence vector of the English text using the coherence vector calculation formula; based on the graph logical coherence vector of the English text, calculate the logical coherence score of the English text using the logical coherence score calculation formula, and output the logical coherence score and comments of the English text.

2. The method for analyzing the logical coherence of English text according to claim 1, characterized in that: The processing steps of the English text logical relation tree generation module are as follows: Starting from page 201; P202 reads the English text to be analyzed; P203 Sentence segmentation of the English text to be analyzed; P204 performs word segmentation on the English text to be analyzed; P205 analyzes the logical relationships of the English text after sentence and word segmentation, and constructs a logical relationship tree for the English text. P206 Output the constructed English text logical relation tree; End of page 207.

3. The method for analyzing the logical coherence of English text according to claim 1, characterized in that: The processing steps of the English text logic diagram conversion module are as follows: Starting from P301; P302 reads the logical relationship tree of English text; P303 Initializes the logical relationship queue; P304 inputs the root node of the logical relation tree of the English text into the logical relation queue; P305 performs a dequeue operation on the head node of the logical relation queue and sets it as the current node; P306 checks if the root node of the first subtree of the current node is empty; otherwise, proceed to operation P307; otherwise, proceed to operation P308. P307 inputs the root node of the first subtree of the current node into the logical relation queue; P308 checks if the root node of the second subtree of the current node is empty; otherwise, proceed to operation P309; otherwise, proceed to operation P310. P309 inputs the root node of the second subtree of the current node into the logical relation queue; P310 checks whether the current node is a leaf node; otherwise, proceed to operation P311; otherwise, proceed to operation P313. P311 sets the core clause of the first subtree of the current node as a node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause. P312 sets the core clause of the second subtree of the current node as the other node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause, and then the operation proceeds to P313. P313 checks if the logical relation queue is empty; otherwise, proceed to P305; otherwise, proceed to P314. Output the logical relationship diagram of the English text on page 314; End of P315.

4. The method for analyzing the logical coherence of English text according to claim 1, characterized in that: The processing steps of the English text logical coherence analysis module are as follows: Starting from P401; P402 Read the logical relationship diagram of English text; P403 Calculate the adjacency grid of the English text logical relationship graph according to the adjacency grid formula (1); Page 404 uses feature vectors to represent the nodes of the logical relationship graph of English text. P405 Calculate the feature grid of the English text logic graph according to the feature grid formula (2); P406 Calculate the coherence vector of the English text logic diagram according to the formula (3) for the coherence vector of the English text logic diagram; P407 Calculate the logical coherence score of English text according to the English text logical coherence score formula (4); P408 generates comments on the logical coherence of English text; End of page 409; Formula for calculating the adjacency grid in an English text logic graph: In formula (1), i and j represent the i-th node and the j-th node in the English text logical relationship graph, respectively. The positions of the i-th node and the j-th node in the adjacency grid of the English text logical relationship graph are represented by i and j. When there is an edge between node i and node j, the corresponding value in the adjacency grid of the English text logical relationship graph is 1, otherwise it is 0. Formula for calculating the feature grid of the logical relationship graph in English text: In formula (2), n represents the total number of nodes in the logical relationship graph of the English text, w represents the word order information of the node, and m represents the total number of words in the node. Formula for calculating the coherence vector of a logical relationship diagram in English text: In formula (3), k represents the kth row corresponding to the kth node in the adjacency grid and feature grid of the English text logical relationship graph, t represents the total number of nodes in the English text logical relationship graph, the adjacency grid of the English text logical relationship graph is obtained by formula (1), and the feature grid of the English text logical relationship graph is obtained by formula (2). Formula for calculating the logical coherence score of English text: In formula (4), r represents the r-th number in the coherent vector of the English text logic diagram, p represents the total dimension of the coherent vector of the English text logic diagram, and e represents the base of the natural logarithm function. The coherent vector of the English text logic diagram is calculated by formula (3).

Images