CN114117054B - Student end-of-term evaluation method, system, device and storage medium based on personalized words - Google Patents

Abstract

The invention discloses a student end-of-term evaluation method, a system, a device and a storage medium based on individual words, which can be widely applied to the technical field of big data processing. According to the invention, the individual descriptive words are mined in the Chinese word vector library according to the acquired individual seed words to form an individual description word library, all the individual descriptive words in the individual description word library are subjected to cluster analysis, the cluster type of each individual descriptive word is determined, and the distribution probability of the individual descriptive words in the cluster type is determined, so that when a teacher performs student end-of-term evaluation, after the individual descriptive words serving as current student end-of-term evaluation information are determined in the cluster type through a preset end-of-term evaluation platform, the distribution probability corresponding to the student individual evaluation words can be automatically determined, and a student end-of-term evaluation graph is generated according to the student individual evaluation words and the distribution probability corresponding to the student individual evaluation words, thereby effectively improving the comprehensiveness and flexibility of the teacher in evaluating the students.

Description

Student end-of-term evaluation method, system, device and storage medium based on personalized words

Technical Field

The invention relates to the technical field of big data processing, in particular to a student end-of-term evaluation method, a student end-of-term evaluation system, a student end-of-term evaluation device and a storage medium based on individual words.

Background

The key of the end-of-term evaluation is not whether to write the student's shortcomings, but whether to fit the student's reality. At present, the evaluation given by the end-of-term teacher usually depends on daily observation and subjective feeling of students to write comments, and lacks of standard grasp on the comprehensive development of the students, so that written comments are lopsided, and the comprehensive quality of the students cannot be comprehensively reflected. And a relatively generalized conclusion is often given, so that the comments are uniform, the personalized features of the students cannot be reflected, and the pertinence and the flexibility are lacked.

Disclosure of Invention

The present invention is directed to solving at least one of the problems of the prior art. Therefore, the invention provides a student end-of-term evaluation method, a system, a device and a storage medium based on individual words, which can effectively improve the comprehensiveness and flexibility of evaluation of students at the end of term.

On one hand, the embodiment of the invention provides a student end-of-term evaluation method based on personalized words, which comprises the following steps:

obtaining individual seed words;

determining a Chinese word vector library meeting a first preset requirement;

individual description words are mined in the Chinese word vector library according to the individual seed words to obtain an individual description word library;

performing cluster analysis on all the individual descriptor in the individual descriptor library, and determining a cluster category to which each individual descriptor belongs and the distribution probability of the individual descriptor in the cluster category;

acquiring individual descriptors in the cluster category in the end-of-term evaluation platform as student individual evaluation words of student end-of-term evaluation information;

determining the distribution probability of the individual evaluation words of the students in the corresponding cluster categories;

and generating a student end evaluation graph according to the student individual evaluation words and the distribution probability of the student individual evaluation words in each cluster category.

In some embodiments, the mining an individual descriptor in the chinese word vector library according to the individual seed word to obtain an individual descriptor library includes:

comparing the plurality of Chinese word vector libraries according to the individual seed words and a preset evaluation index;

determining a target word vector library according to a preset application scene and a comparison result;

and obtaining an individual descriptor similar to the individual seed word from the target word vector library, and storing the individual descriptor to an individual descriptor library.

In some embodiments, the comparing the plurality of chinese word vector libraries according to the personalized seed word and a preset evaluation index includes:

acquiring the number of participles contained in the word vectors in the Chinese word vector library;

determining the receiving and recording proportion of the individual seed words in the Chinese word vector libraries;

determining context information of word vectors in the plurality of Chinese word vector libraries according to the individual seed words;

determining the similar meaning words of the individual seed words in the Chinese word vector libraries, and determining the similarity distribution information of the individual seed words and the similar meaning words;

and comparing the plurality of Chinese word vector libraries according to at least one of the word segmentation number, the receiving and recording proportion, the context information or the similarity distribution information.

In some embodiments, after the step of determining the target word vector library according to the preset application scenario and the comparison result, the method further includes the following steps:

and eliminating the individual descriptive words meeting second preset requirements in the target word vector library.

In some embodiments, the obtaining, in the target word vector library, an individual descriptor similar to the individual seed word includes:

obtaining near-meaning words of the individual seed words with preset number from a target word vector library after the individual descriptor meeting a second preset requirement is removed as individual descriptors;

or

Obtaining a near-meaning word of the individual seed word from a target word vector library after the individual descriptor meeting a second preset requirement is removed;

determining the similarity between the individual descriptor and the similar meaning word;

and determining the similar meaning words with the similarity meeting a preset threshold as the individual description words.

In some embodiments, the performing cluster analysis on all the individual descriptors in the individual descriptor library includes:

carrying out Gaussian mixture model clustering on all the individual descriptors in the individual descriptor library;

and analyzing, evaluating, integrating and inducing the clustering result of the Gaussian mixture model according to the contour coefficient.

In some embodiments, the method further comprises the steps of:

obtaining a student viewing request uploaded by the preset end-of-term evaluation platform;

and controlling the preset end-of-term evaluation platform to display a word cloud image or cluster information of the individual student evaluation words corresponding to the student end-of-term evaluation graph according to the student viewing request.

On the other hand, the embodiment of the invention provides a student end-of-term evaluation system based on personalized words, which comprises the following components:

the first acquisition module is used for acquiring the individual seed words;

the mining module is used for determining a Chinese word vector library meeting a first preset requirement and mining an individual description word in the Chinese word vector library according to the individual seed word to obtain an individual description word library;

the clustering module is used for carrying out clustering analysis on all the individual descriptors in the individual descriptor library, and determining the clustering class to which each individual descriptor belongs and the distribution probability of the individual descriptors in the clustering class;

the second acquisition module is used for acquiring the individual descriptors in the cluster categories in the end-of-term evaluation platform as student individual evaluation terms of the student end-of-term evaluation information;

the determining module is used for determining the distribution probability of the student individual evaluation words in the corresponding clustering categories;

and the generation module is used for generating a student end-of-term evaluation graph according to the student individual evaluation words and the distribution probability of the student individual evaluation words in each cluster category.

On the other hand, the embodiment of the invention provides a student end-of-term evaluation device based on personalized words, which comprises:

at least one memory for storing a program;

at least one processor for loading the program to execute the student end-of-term assessment method based on the personality words.

In another aspect, an embodiment of the present invention provides a storage medium, in which a computer-executable program is stored, and the computer-executable program is executed by a processor to implement the end-of-term student evaluation method based on personalized words.

The student end-of-term evaluation method based on the individual words provided by the embodiment of the invention has the following beneficial effects:

in the embodiment, the individual description words are mined in the Chinese word vector library according to the acquired individual seed words to form an individual description word library, all the individual description words in the individual description word library are subjected to cluster analysis, the cluster type of each individual description word is determined, and the distribution probability of the individual description words in the cluster type is determined, so that when a teacher performs student end evaluation, after the individual description words serving as current student end evaluation information are determined in the cluster type through a preset end-of-term evaluation platform, the corresponding distribution probability of the student individual evaluation words in each cluster type can be automatically determined, and a student end-of-term evaluation graph is generated according to the corresponding distribution probabilities of the student individual evaluation words and the student individual evaluation words, so that the comprehensiveness and flexibility of the teacher in evaluating the students are effectively improved.

Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.

Drawings

The invention is further described with reference to the following figures and examples, in which:

fig. 1 is a flowchart of a student end-of-term evaluation method based on personalized words according to an embodiment of the present invention;

fig. 2 is a flow chart of acquiring a personal description lexicon according to an embodiment of the present invention;

FIG. 3 is a diagram illustrating a method for searching a personal descriptor according to an embodiment of the present invention;

FIG. 4 is a probability distribution diagram of a type of personality descriptor in accordance with an embodiment of the present invention;

fig. 5 is a radar schematic diagram corresponding to the student end-of-term evaluation chart in the embodiment of the present invention.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.

In the description of the present invention, it should be understood that the orientation or positional relationship referred to in the description of the orientation, such as the upper, lower, front, rear, left, right, etc., is based on the orientation or positional relationship shown in the drawings, and is only for convenience of description and simplification of description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus, should not be construed as limiting the present invention.

In the description of the present invention, the meaning of a plurality is one or more, the meaning of a plurality is two or more, and the above, below, exceeding, etc. are understood as excluding the present numbers, and the above, below, within, etc. are understood as including the present numbers. If the first and second are described for the purpose of distinguishing technical features, they are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.

In the description of the present invention, unless otherwise explicitly limited, terms such as arrangement, installation, connection and the like should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meanings of the above terms in the present invention in combination with the specific contents of the technical solutions.

In the description of the present invention, reference to the description of the terms "one embodiment," "some embodiments," "an illustrative embodiment," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

In the teaching process, the key of the teacher's evaluation on the students is not to write the defects of the students, but to evaluate in combination with the actual conditions of the students. The assessment is moderate, cannot be affirmed, expressed, exaggerated and denied at one time, is based on facts, avoids the evaluation of holes and lack of individual characteristics, reflects the individual characteristics of the students, can promote the healthy growth and development of the students, and plays a role in encouraging and guiding the students. At present, the evaluation of students by teachers mainly has the following problems:

first, objectivity is insufficient, evaluation criteria are dislocated: the term end comment generally points out the practical situation and some defects of the students, so the language should be objective, clear, concise and powerful, and ask the students to correctly recognize the problems and to clearly understand the direction of the efforts. However, at present, many teachers write comments by virtue of daily observation and subjective feelings of students, and lack of standard grasp on the comprehensive development of the students, so that the writing of the price words is lopsided, and the comment content cannot comprehensively reflect the comprehensive quality of the students. Meanwhile, the written expression of the comment is not complete enough, so that the teacher has less serious avoidance on the words for operating the comment, and the practical situation of the student is difficult to embody objectively.

Secondly, the individuation is insufficient, and the homogenization is serious: the personalized language in the comment can not only enhance the self-confidence of students, but also stimulate the students to give strong motivation to give forward, and because some teachers pay less attention to the students at present, more generalized conclusions are often given, so that the comment content is uniform, the personalized characteristics of the students cannot be reflected, and the pertinence and the flexibility are lacked. Thus, the end comment with homogenization problem restricts the effective exertion of the education evaluation function, and the self acceptance and the improvement of the self efficiency sense of the students cannot exert due suggestion and guidance.

Thirdly, improper language expression: the text expression ability of part of teachers is insufficient, and the language expression is stiff and rigid, so that students cannot really make sure the actual conditions of the teachers. When teachers write comments, the phenomena of monotony, dryness and verbosity need to be avoided. Still other comments have the problem of excessive incite of language or oral type motivation flooding and disaster, neither conform to the age characteristics of students, nor are too far away from the students' actual abilities.

Based on the above, the embodiment of the invention provides a student end-of-term evaluation method, a system, a device and a storage medium based on personalized words. Specifically, in the embodiment, the individual descriptor is mined in the Chinese word vector library according to the acquired individual seed words to form an individual descriptor library, all the individual descriptors in the individual descriptor library are subjected to cluster analysis, the cluster type to which each individual descriptor belongs and the distribution probability of the individual descriptors in the cluster type are determined, so that when a teacher performs student end-of-term evaluation, after the individual descriptors serving as current student end-of-term evaluation information are determined in the cluster type through a preset end-of-term evaluation platform, the distribution probability corresponding to the student individual evaluation words can be automatically determined, and a student end-of-term evaluation graph is generated according to the student individual evaluation words and the distribution probability corresponding to the student individual evaluation words, so that the comprehensiveness and flexibility of the teacher in evaluating students are effectively improved.

The invention is further illustrated in the following description with reference to the accompanying drawings:

referring to fig. 1, an embodiment of the present invention provides a student end-of-term evaluation method based on an individual word, and the embodiment may be applied to a background processor corresponding to an automatic evaluation platform of a teacher for students, and may also be applied to a server. During the application process, the background processor or the server can perform data interaction with each application terminal.

Specifically, the present embodiment includes the following steps:

and S11, obtaining the individual seed words.

In this embodiment, a small number of individual descriptors for evaluating students may be constructed by referring to a dictionary, association, or the like, and these small number of individual descriptors may be used as individual seed words. These personality seed words include not only recognition evaluation words for students but also derogation evaluation words for students. The method can be used for screening and developing the association of positive and negative words based on the existing Chinese personality word research, for example, if the personality word is 'generous', the 'small atmosphere' of the 'generous' negative word can be found by looking up a dictionary and the like. The words obtained by reference or association should contain some behavior evaluation words suitable for students as much as possible. In this embodiment, the personality descriptor is a natural language vocabulary describing the individual person. In personalised psychology, generally called personality descriptors, it describes a general tendency of some behaviour of people, such as modesty, familiarity, irritability, irresistance and so on. In the end evaluation, the teacher often describes the behavior and the personality development of the student in the period by using the words, and the embodiment is expected to describe and model the behavior tendency of the student in the period by using the personality description words instead of defining each student in a uniform text form.

And S12, determining the Chinese word vector library meeting the first preset requirement, and mining individual description words in the Chinese word vector library according to the individual seed words to obtain an individual description word library.

In this embodiment, the first preset requirement may be set according to the characteristics of whether the first preset requirement is applicable to the end-of-term evaluation of the student. The Chinese word vector library comprises a plurality of word vector libraries which meet a first preset requirement. For example, a known Chinese word vector library such as an Tencent word vector library, a Wikipedia word vector library, a word vector library corresponding to the Hopkins, and the like is obtained. Specifically, as shown in fig. 2, the personality description lexicon can be obtained by the following ways:

and S21, comparing the Chinese word vector libraries according to the individual seed words and the preset evaluation indexes.

In this embodiment, the number of participles contained in a word vector in a plurality of chinese word vector libraries may be obtained, and the plurality of chinese word vector libraries may be compared according to the preset evaluation index of the number of participles; the method comprises the steps of firstly determining the inclusion proportion of individual seed words in a plurality of Chinese word vector libraries, and then comparing the plurality of Chinese word vector libraries according to the inclusion proportion which is a preset evaluation index; the context information of word vectors in a plurality of Chinese word vector libraries can be determined according to the individual seed words, and then the plurality of Chinese word vector libraries are compared according to the preset evaluation index of the context information. For example, a word vector library based on Tencent social language training is more consistent with the language habit of people in daily life, while a word vector library based on Wikipedia linguistic data is closer to a language of popular science type and is not suitable for being used as a linguistic data for describing people in daily context. The similar meaning words mentioned in the step refer to words with similar contexts described in a word vector space, for example, the tolerant similar meaning words may include inclusion or may also include non-tolerance, inspiration and the like, because the context descriptions of the tolerant and inspiration in the natural language are always similar, so the similar meaning words in the step include both possible positive descriptions and negative personal word evaluation descriptions which are opposite but consistent in description; and after determining the similarity distribution information of the individual seed words and the similar meaning words, comparing the plurality of Chinese word vector libraries according to the preset evaluation index of the similarity distribution information. In addition, the Chinese word vector libraries can be compared according to at least two preset evaluation indexes of the word segmentation number, the receiving and recording proportion, the context information or the similarity distribution information, and the specific comparison mode can be adjusted according to the actual situation.

And S22, determining a target word vector library according to the preset application scene and the comparison result.

In this embodiment, after the comparison of the word vector libraries, a relatively suitable specific application scenario may be selected, for example, students evaluating different age groups may select different word vector libraries as a target word vector library, and individual descriptors meeting a second preset requirement in the target word vector library are removed. The method can be used for eliminating rare individual seed words which are not included in a word vector library, for example, individual descriptors such as 'ship support in slaughtered belly' may not be included in corresponding word vector participles, and then the individual descriptor words need to be eliminated.

And S23, obtaining the individual descriptor similar to the individual seed word from the target word vector library, and storing the individual descriptor into an individual descriptor library.

In this embodiment, when the number of the individual seed words is small, in order to obtain more similar individual descriptors as much as possible, the near-synonyms of the individual seed words in the preset number may be obtained as the individual descriptors in the target word vector library from which the individual descriptors meeting the second preset requirement are removed. Specifically, 10 to 20 similar meaning words near the personality seed word can be acquired as the preliminary personality descriptor and stored in the personality descriptor bank. For example, 20 synonyms closest to the "forgiveness" are searched as personality descriptor regardless of the similarity, and saved to the personality descriptor repository. When the number of the collected individual seed words is large, in order to more accurately collect the similar meaning words and the vectors thereof, the similar meaning words of the individual seed words can be firstly obtained in the target word vector library after the individual descriptor meeting the second preset requirement is removed, then the similarity between the individual descriptor and the similar meaning words is determined, and then the similar meaning words meeting the preset threshold value in the similarity are determined to be used as the individual descriptor. For example, a preset threshold may be set to be approximately equal to 1, the closer to 1, the higher the representing requirement is, when retrieving a near-meaning word of the individual seed word, whether the similarity between the near-meaning word and the seed word is greater than the preset threshold is determined, and when the similarity is greater than the preset threshold, the individual seed word is included as the individual word bank of the student. And then, continuously and iteratively searching the similar words of the found individual description word similar words, and continuously judging whether the similarity is greater than a preset threshold value or not, and jumping out the work of traversing the word vector until the similarity of all the similar words does not meet the preset threshold value any more. Representing that the preliminary personality lexicon has been constructed. As shown in fig. 3, assuming that the preset threshold is 0.85, inputting the seed word "tolerance" into the Tencent word vector library, searching for words that contain the word that is close to the "tolerance" vector until the similarity is less than 0.85, stopping searching, determining the similarity to the "tolerance" word from the found close words, if the similarity is still greater than 0.85, continuing to contain the words until the similarity to the "tolerance" vector is all less than 0.85, stopping containing the word that describes the individual meaning. By the method, the word quantity expansion can be performed on the personality description word bank.

In this embodiment, after the personality descriptor library is obtained, the personality descriptors in the personality descriptor library are further screened, and some personality descriptors which are too negative and disambiguated and are not suitable for describing students, such as descriptors of loving virtues, fool foolproof and dead brains, are removed. After the word vectors in the character description word library are removed, the residual character description words, the seed words and the word vectors in the vector library are stored in the character description word library together. The subsequent steps are processed by the information in the personality description library. Where Word embedding is a general term for a set of language modeling and feature learning techniques in Word embedded Natural Language Processing (NLP), where words or phrases from a vocabulary are mapped to vectors of real numbers. Conceptually, it involves mathematical embedding from a one-dimensional space of each word to a continuous vector space with lower dimensions. Methods of generating such mappings include neural networks, dimensionality reduction of word co-occurrence matrices, probabilistic models, interpretable knowledge base methods, and context for explicit representation of word occurrences of terms. Word and phrase embedding, when used as the underlying input representation, has been shown to improve the performance of NLP tasks, such as parsing and sentiment analysis. Specifically, a word vector is a method of representing words with vectors trained from a large amount of text based on its contextual information. As simple word vector assumptions, from the human natural language, the meaning context of similar words may be consistent, such as "thirst" and "hunger":

i are now very thirsty.

I are now hungry.

The vectors of "thirst" and "hunger" above, which should be similar in vector space, i.e. synonyms, can be represented based on different algorithms. But training with a very large number of samples is required to accurately represent the near-sense relationship between the two words, and the characteristics of different word vectors are different. For example, the word vector may have different dimensions, different sizes of the contained segments, different amounts of the rich context information, and the like, and may contain context information, semantic information, emotion information, and the like due to the interpretability of the neural network. In this embodiment, for example, the Tencent word vector data with a good Chinese recognized training effect is taken as an example, the tolerant word vector may be represented as a vector with a length of 200 dimensions. It is convenient to locate a word in the word vector space that has been trained to converge and find his particular number of near-synonyms. If "forgiveness" is entered, locating where it is in the Tencent word vector, and finding the 10 word vectors closest to him, then "understand, forgiveness treatment, understand forgiveness, require forgiveness, contain, forgiveness other person, forgiveness admission, forgiveness heart and contain" will be returned. Therefore, on the basis of the specific individual seed words, the near meaning words are searched in the appropriate word vector space to obtain the student evaluation individual words with enough coverage.

S13, performing cluster analysis on all the individual descriptors in the individual descriptor library, and determining the cluster category to which each individual descriptor belongs and the distribution probability of the individual descriptors in the cluster category.

In this embodiment, after the gaussian mixture model clustering is performed on all the individual descriptors in the individual descriptor library, the gaussian mixture model clustering result is analyzed, evaluated, integrated and induced according to the contour coefficient. Specifically, a gaussian mixture model GMM is used to perform cluster analysis on the obtained individual descriptive words, and the individual words in the same cluster describe the evaluation modes of the same aspect in the natural language in the chinese context, for example, proud and humble are divided into the same cluster, which are all dimensions describing whether a person is humorous or not. Taking table 1 as an example, the main information described by each category after more than 10000 individual descriptors are clustered into six categories is shown. Wherein the negative description in table 1 does not refer to a negative description, but a positive negative, without emotional colors.

TABLE 1

Categories	Number of words of description	Description of the front	Negative description
				A	3193	Low tone, wide appearance, hardship, thrill and faith	Thinness, impatience, irritability, selfishness, and leisure love
B	2581	Impartial, responsible, collective, rational and active	Closed, perceptual, passive, pessimistic, personal senses
				C	2546	Open thought, humour, leadership and dry practice	Laziness, conservation, spreading and spreading,Dragging tool
D	2613	Positive, friendly, cool, sunshine and sharp	Sensitivity, strange autism, carelessness and impatience
				E	3231	Straighten, modest, courteous, acquainted, patriotic, noble, law-keen	Proud, disorder of law, waste, shallow skin and lugling
F	1504	Simple, good at the same heart, benefiting him, vitality and good at the interpersonal	Slow, traitory, doubtful, sloppy, indifferent

After the individual descriptors are clustered, the number of possible clusters in the cluster is continuously tried through the contour coefficient cluster evaluation index to obtain the most appropriate clustering result of the individual descriptors, and the cluster category to which each individual descriptor belongs and the probability distribution of the cluster category are stored. In this embodiment, the contour coefficient is an evaluation index for evaluating the cluster analysis effect. By analyzing different clustering numbers, the contour coefficient reflects different variation trends. The result of a particular cluster may be determined from the change in the contour coefficients. The value of the contour coefficient is [ -1,1], and the closer to 1, the better the clustering effect is. The Gaussian mixture model GMM refers to a clustering analysis method. It clusters a group of samples into a certain number of clusters (classes) and, in addition to returning to which class each sample should belong, it also returns the probability distribution of the other classes to which this sample belongs. That is, if a sample is subjected to GMM clustering, it can obtain not only the result of the possible cluster to which the word belongs, but also the distribution of other probabilities to which the word belongs.

For example, the optimal clustering result of more than 10000 individual descriptors in the word flight vector according to the contour coefficient is 6 in table 1, which represents that 10000 individual descriptors collected can be clustered into six categories, wherein the description is that the behavior of people has six aspects in the context of chinese. For example, A, B, C, D, E and F in table 1 correspond to six behavior description forms obtained by clustering 10000 individual description words, which are referred to as personality traits in psychology, and as shown in fig. 4, the probability of the first class of words distributed according to the word can be represented, and the closer to the center, the higher the probability of the word being classified into the first class, and the closer to the edge, the word may be clustered into other clusters. For example, the term "hardship and frugal saving" means that the probability of the behavior tendency in the first category is 0.7 and the probability of the behavior tendency in the fourth category is 0.2%, which means that the term "hardship and frugal saving" can represent the behavior of a certain person in the first category of behavior tendency and can also partially represent the behavior of a certain student in the fourth category of behavior tendency.

For another example, the word "pride" is also clustered as the first category, and the similar words also include: blowing cattle, not proud, playing heart and eye, 26688, \39580, passenger qi and the like, and the description of the same behavior aspect of the class comprises positive and negative descriptions. For proud, although it is clustered as the first class, as shown in table 2, GMM will return its classification "probability" in other five classes, and count their probability distribution rules according to the selected individual descriptive words to obtain a comprehensive quantitative evaluation model of a certain school period of the corresponding student.

TABLE 2

Class 1 behavioral trends

Class

2 behavioral trends

Class

3 behavioral trends

Class

4 behavioral trends

Category

5 behavior trends

Category

6 behavior trends

Sum of six kinds of probabilities

Individual descriptor

0.751903774

1.85E-06

0.180451351

0.067595662

4.74E-05

3.49E-10

1

Proud

And S14, acquiring the individual descriptors in the cluster categories in the end-of-term evaluation platform as the individual evaluation terms of the students of the end-of-term evaluation information.

In the implementation, when a teacher needs to evaluate students, the individual descriptors serving as the current student end-of-term evaluation information can be selected from the cluster type on the preset end-of-term evaluation platform, so that the teacher can better select the individual descriptors according with the practical conditions of the students to evaluate, and the flexibility and the comprehensiveness of the evaluation process are improved. For example, a teacher logs in a built end-of-term evaluation platform, a background of the platform stores the individual description word bank obtained in the above steps, a front end of the platform is displayed as a word cloud of the individual word bank, the teacher can conveniently screen and submit individual words, the teacher selects the words according to the behavior of each student in the period, and then the background completes modeling of end-of-term evaluation on the behavior of the learner in the period based on subsequent steps. In different individual word clustering results, the teacher selects individual words on the platform to evaluate the academic behaviors and personality of the students, wherein positive individual word evaluation such as modesty, familiarity and the like and negative individual word evaluation such as proud, high coldness and the like are included.

And S15, determining the distribution probability of the individual evaluation words of the students in the corresponding cluster categories.

In this embodiment, after obtaining the student individual evaluation words of the current student uploaded by the platform, the distribution probability corresponding to the student individual evaluation words is determined from the individual description word library, so as to facilitate the subsequent modeling representation of the end-of-term evaluation of the student. Specifically, assuming that the teacher selects the evaluation description of the study period of student a as shown in table 3, the corresponding distribution probability is also shown in table 3:

TABLE 3

Class 1 behavioral trends	Class	2 behavioral trends	Class	3 behavioral trends	Class	4 behavioral trends	Category	5 behavior trends	Category	6 behavior trends	Sum of six kinds of probabilities	Student individual evaluation word
													1.81E-06	0.001629	0.012252	0.00019	0.079064	0.906863	1	Wide-capacity others
5.69E-05	0.000488	0.068273	0.005522	0.005891	0.919769	1	Trusting others
								0.996008107	0.002564216	6.76E-08	0.000246511	0.001181098	4.43E-28	1	Lovely pure
0.000307	7.00E-05	0.007966	1.46E-05	0.442783	0.548859	1	Pleased to help others
								2.11E-06	0.000966	0.004842	1.63E-05	0.287912	0.706261	1	Thanksgiving teacher
0.915901963	1.32E-10	0.083562634	7.93E-05	0.000381944	7.41E-05	1	Without greedy
								0.999823216	5.10E-05	8.40E-10	2.27E-05	0.000103028	1.36E-36	1	Quiet
0.000109	5.47E-07	0.208355	0.086791	0.704743	3.78E-07	1	Comparing inward
								0.751903774	1.85E-06	0.180451351	0.067595662	4.74E-05	3.49E-10	1	Proud
0.337355186	7.36E-08	0.327757585	0.308137655	1.76E-06	0.026747741	1	Play spleen qi
								0.517927761	2.75E-05	0.319607268	0.162265903	0.000171532	1.55E-08	1	Acute manifestation
0.410854257	0.00052711	0.11027881	0.057352882	0.138389066	0.282597661	1	The school date is as follows: behavioral tendency modeling of A students

In the probability calculation process in table 3, the probability of each word of the classmate belonging to the category is selected by the teacher, and the average probability of each category is calculated as the model of the classmate schooling behavior. For student A, the behavior tendency modeling calculation method is to calculate the average value of the probabilities of each column of behavior tendency clusters. The sum of the probability distribution values for the behavior tendency modeling is also 1, the last row in the table. For the evaluation of a plurality of students, after the above processing procedure is carried out on each student, the probability distribution of each student in each classification and the evaluated words are saved as the behavior evaluation model of the scholarly.

And S16, generating a student end evaluation graph according to the student individual evaluation words and the distribution probability of the student individual evaluation words in each cluster category.

In the embodiment, after obtaining the probability distribution of each student in each individual personality descriptor category, for example, the comprehensive probability distribution of the student a in terms of six personality descriptors in the last row in table 3 is converted into a radar map as shown in fig. 5 according to the probability distribution of the six categories, and is used as the student end evaluation map of the current student in the current school period. The probability distribution of behavior of each student is converted into a radar form as shown in fig. 5. After the students log in the platform, the radar schematic diagram is displayed on a corresponding user interface, so that the students can better know the evaluation of the teachers on the students. If the students want to know more detailed specific evaluation of the self-school date, the students can send the student check requests through the current interface, and the server can control the preset end-of-term evaluation platform to display the word cloud image or the clustering information of the individual student evaluation words corresponding to the student end-of-term evaluation image according to the student check requests.

In conclusion, the embodiment enables the teacher to more conveniently select and evaluate the behavior of each school term of the student through the existing individual descriptive words, and does not need to organize and evaluate the expressions of the excessive adjustment of the term of the student, so that the evaluation of the school term of the student can be completed, and the behavior tendency and the idiosyncratic composition of the student can be quantitatively analyzed and more known. Meanwhile, the evaluation of the end of the period of different scholars can be compared. And observing whether the students have psychological problems such as behavior tendency mutation and the like in the learning and growth processes. For example, the first class behavior tendency of the student A evaluated by the same teacher in the first school period is obvious, the second school period suddenly shows that the first class behavior tendency is atrophied, and the second class behavior tendency suddenly shows that the psychological state of the student is changed and needs to intervene in time.

The embodiment of the invention provides a student end-of-term evaluation system based on personalized words, which comprises:

the first acquisition module is used for acquiring the individual seed words;

the mining module is used for determining a Chinese word vector library meeting a first preset requirement and mining individual description words in the Chinese word vector library according to the individual seed words to obtain an individual description word library;

the clustering module is used for carrying out clustering analysis on all the individual descriptors in the individual descriptor library, and determining the cluster category to which each individual descriptor belongs and the distribution probability of the individual descriptors in the cluster category;

the second acquisition module is used for acquiring the individual descriptors in the cluster categories in the end-of-term evaluation platform as student individual evaluation terms of the student end-of-term evaluation information;

the determining module is used for determining the distribution probability of the individual evaluation words of the students in the corresponding clustering categories;

and the generation module is used for generating a student end evaluation graph according to the student individual evaluation words and the distribution probability of the student individual evaluation words in each cluster category.

The content of the embodiment of the method of the invention is all applicable to the embodiment of the system, the function of the embodiment of the system is the same as the embodiment of the method, and the beneficial effect achieved by the embodiment of the system is the same as the beneficial effect achieved by the method.

The embodiment of the invention provides a student end-of-term evaluation device based on personalized words, which comprises:

at least one memory for storing a program;

at least one processor configured to load the program to perform the end-of-term student evaluation method based on the personality words shown in fig. 1.

The content of the method embodiment of the present invention is applicable to the apparatus embodiment, the functions specifically implemented by the apparatus embodiment are the same as those of the method embodiment, and the beneficial effects achieved by the apparatus embodiment are also the same as those achieved by the method.

An embodiment of the present invention provides a storage medium in which a computer-executable program is stored, and the computer-executable program is executed by a processor to implement the end-of-term student evaluation method based on personalized words shown in fig. 1.

The embodiment of the invention also discloses a computer program product or a computer program, which comprises computer instructions, and the computer instructions are stored in a computer readable storage medium. The computer instructions may be read by a processor of a computer device from a computer-readable storage medium, and the computer instructions executed by the processor cause the computer device to perform the end-of-term student evaluation method based on personality words shown in fig. 1.

The embodiments of the present invention have been described in detail with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention. Furthermore, the embodiments of the present invention and the features of the embodiments may be combined with each other without conflict.

Claims (10)

1. A student end-of-term evaluation method based on personalized words is characterized by comprising the following steps:

obtaining individual seed words;

determining a Chinese word vector library meeting a first preset requirement;

individual descriptors are mined in the Chinese word vector library according to the individual seed words to obtain an individual description word library, the individual description word library is a target Chinese word vector library suitable for the characteristics of student end evaluation, and the individual descriptors do not include negative and depreciated descriptors;

performing cluster analysis on all the individual descriptors in the individual descriptor library, and determining a cluster category to which each individual descriptor belongs and the distribution probability of the individual descriptors in the cluster category;

obtaining individual descriptors in the cluster categories selected in an end-of-term evaluation platform after a teacher logs in the end-of-term evaluation platform as student individual evaluation terms of student end-of-term evaluation information;

determining the distribution probability of the student individual evaluation words in the corresponding cluster categories;

and generating a student end evaluation graph according to the student individual evaluation words and the distribution probability of the student individual evaluation words in each cluster category.

2. The student end-of-term evaluation method based on personalized words according to claim 1, wherein the mining of personalized descriptors in the Chinese word vector library according to the personalized seed words to obtain a personalized descriptor library comprises:

comparing a plurality of Chinese word vector libraries according to the individual seed words and preset evaluation indexes;

determining a target word vector library according to a preset application scene and a comparison result;

and obtaining an individual descriptor similar to the individual seed word from the target word vector library, and storing the individual descriptor to an individual descriptor library.

3. The student end-of-term evaluation method based on personalized words according to claim 2, wherein comparing a plurality of Chinese word vector libraries according to the personalized seed words and preset evaluation indexes comprises:

acquiring the number of participles contained in the word vectors in the Chinese word vector library;

determining the receiving and recording proportion of the individual seed words in the Chinese word vector libraries;

determining context information of word vectors in the plurality of Chinese word vector libraries according to the individual seed words;

determining the similar meaning words of the individual seed words in the Chinese word vector libraries, and determining the similarity distribution information of the individual seed words and the similar meaning words;

and comparing the plurality of Chinese word vector libraries according to at least one of the word segmentation number, the receiving and recording proportion, the context information or the similarity distribution information.

4. The student end-of-term evaluation method based on personalized words according to claim 2, wherein after the step of determining the target word vector library according to the preset application scenario and the comparison result, the method further comprises the following steps:

and eliminating the individual descriptive words meeting second preset requirements in the target word vector library.

5. The method as claimed in claim 4, wherein the step of obtaining the individual descriptor similar to the individual seed word from the target word vector database comprises:

obtaining near-meaning words of the individual seed words with preset number from a target word vector library after the individual descriptor meeting a second preset requirement is removed as individual descriptors;

or alternatively

Obtaining a near-meaning word of the individual seed word from a target word vector library after the individual descriptor meeting a second preset requirement is removed;

determining the similarity of the individual descriptor and the similar word;

and determining the similar meaning words with the similarity meeting a preset threshold as the individual description words.

6. The student end-of-term evaluation method based on personal words as claimed in claim 1, wherein the clustering analysis of all personal descriptors in the personal descriptor library comprises:

carrying out Gaussian mixture model clustering on all the individual descriptors in the individual descriptor library;

and analyzing, evaluating, integrating and inducing the clustering result of the Gaussian mixture model according to the contour coefficient.

7. The student end-of-term assessment method based on personalized words according to claim 1, further comprising the steps of:

acquiring a student viewing request uploaded by a preset end-of-term evaluation platform;

and controlling the preset end-of-term evaluation platform to display a word cloud image or cluster information of the individual student evaluation words corresponding to the student end-of-term evaluation graph according to the student viewing request.

8. An end-of-term student evaluation system based on personalized words, comprising:

the first acquisition module is used for acquiring the individual seed words;

the system comprises a mining module, a characteristic description word library and a judgment module, wherein the mining module is used for determining a Chinese word vector library meeting a first preset requirement, mining a characteristic description word in the Chinese word vector library according to the characteristic seed word to obtain the characteristic description word library, the characteristic description word library is a target Chinese word vector library suitable for the end-of-term evaluation characteristics of students, and the characteristic description word does not comprise negative and depreciated description words;

the clustering module is used for carrying out clustering analysis on all the individual descriptors in the individual descriptor library, and determining the clustering class to which each individual descriptor belongs and the distribution probability of the individual descriptors in the clustering class;

the second acquisition module is used for acquiring individual descriptors in the cluster categories selected in the end-of-term evaluation platform after a teacher logs in the end-of-term evaluation platform as student individual evaluation terms of student end-of-term evaluation information;

the determining module is used for determining the distribution probability of the student individual evaluation words in the corresponding clustering categories;

and the generation module is used for generating a student end evaluation graph according to the student individual evaluation words and the distribution probability of the student individual evaluation words in each cluster category.

9. An end-of-term student evaluation device based on personalized words, comprising:

at least one memory for storing a program;

at least one processor configured to load the program to perform the method for student end-of-term assessment based on personalized words according to any of claims 1 to 7.

10. A storage medium having stored therein a computer-executable program which, when executed by a processor, is adapted to implement a student end-of-term assessment method based on personality words according to any one of claims 1-7.

Images