WO2018161917A1

WO2018161917A1 - Intelligent scoring method and apparatus, computer device, and computer-readable medium

Info

Publication number: WO2018161917A1
Application number: PCT/CN2018/078278
Authority: WO
Inventors: 李培帅; 杨磊; 张虎; 杨延超
Original assignee: 百度在线网络技术（北京）有限公司
Priority date: 2017-03-07
Filing date: 2018-03-07
Publication date: 2018-09-13
Also published as: CN106940788A; CN106940788B

Abstract

The invention provides an intelligent scoring method and apparatus, a computer device, and a computer-readable medium. The method comprises: obtaining a user answer corresponding to an examination question; generating, according to the user answer, a user sentence segment library corresponding to the user answer; and scoring the user answer according to the user sentence segment library, a pre-generated standard sentence segment library corresponding to a standard answer, and a weight of each standard sentence segment in the standard sentence segment library. The technical solution of the invention can intelligently score a user answer to a subjective question, and can eliminate defects in the prior art in which it takes a long time to perform scoring during manual marking of examination papers and labor costs are high, effectively shortening the time spent on scoring and reducing labor costs, thereby greatly improving efficiency of scoring a user answer to a subjective question.

Description

Intelligent scoring method and device, computer device and computer readable medium

The present application claims priority to Chinese Patent Application No. 201710131133.6, entitled "Intelligent Scoring Method and Apparatus, Computer Equipment, and Computer-Readable Medium", filed on March 7, 2017.

Technical field

The present invention relates to the field of intelligent computer device technologies, and in particular, to an intelligent scoring method and apparatus, a computer device, and a computer readable medium.

Background technique

In the training, examination, education and other systems, in order to assess the learning situation of the students, it is usually necessary to take the examinations for the students and score the examination results to understand the mastery of the students' knowledge of the learning.

Usually, in order to effectively evaluate the students, the test questions can be divided into subjective questions and objective questions. Among them, the objective questions mostly adopt the method of multiple-choice questions, and the user selects one or more answers from multiple choices according to the questions of the questions. Since the answer to the objective question is fixed, it is very convenient to use the computer to score the objective problem, thereby avoiding the old manual marking method, which can shorten the marking time, save the labor cost of marking, and improve the efficiency of marking. For subjective questions, students usually use the way of discussion to answer questions in the questions. In the process of answering subjective questions, students usually need to play their own way and solve the questions in the questions according to their own way of thinking. Therefore, the answer to the subjective question is only a reference, not an absolute standard, and it is impossible to use a computer sentence to judge how many points the student's answer should be. Therefore, in the prior art, the subjective question is usually scored by means of an older scoring method, that is, a manual scoring method.

However, the method of scoring the answers of the subjective questions by the relatively old manual scoring method adopted by the prior art not only has a long scoring time but also a high labor cost. Therefore, the existing method of using manual scoring is subjective. The answer is less efficient to score.

Summary of the invention

The invention provides an intelligent scoring method and device, a computer device and a computer readable medium for improving the efficiency of scoring an answer of a subjective question.

The invention provides a smart scoring method, the method comprising:

Obtain the user answer corresponding to the question;

Generating a user word segment corresponding to the user answer according to the user answer;

The user answers are scored according to the user segmentation database, the standard word segment corresponding to the pre-generated standard answer, and the weights of the standard word segments in the standard word segment.

Further optionally, in the method as described above, the obtaining is based on the weight of the standard word segment corresponding to the user-based word bank, the pre-generated standard answer, and the standard word segmentation in the standard word segment. Before the standard answer is a score for scoring the user answer, the method further includes:

Obtaining the test question and the standard answer corresponding to the test question;

Generating a standard word segment corresponding to the standard answer according to the standard answer;

A weight is set for each of the standard word segments in the standard word segmentation.

Further, optionally, in the method as described above, before the user segmentation corresponding to the user answer is generated according to the user answer, the method further includes:

Obtaining the questions used by the user to make the user's answer;

Detecting and determining that the test questions used by the user to make the user answer are consistent with the test questions corresponding to the standard answer.

Further optionally, in the method as described above, setting weights for each of the standard word segments in the standard word segmentation library includes:

Counting the frequency of occurrence of each of the standard participles in the standard answer, and setting weights for each of the standard participles according to the frequency of occurrence of each of the standard participles in the standard answer, so that the standard participle with high frequency of occurrence occurs The weight of the standard is greater than the weight of the standard participle with a low frequency of occurrence;

Or displaying each of the standard participles for the judge to set a weight for each of the standard participles; and receiving the weight of each of the standard participles input by the judger through the human interface module.

Further optionally, in the method as described above, generating a user word segment corresponding to the user answer according to the user answer, specifically:

Segmenting the user's answer and obtaining a number of original user word segments;

Filtering the plurality of original user participles according to each filter word in the preset filter word lexicon to obtain a plurality of the user participles;

Combining a plurality of the user participles to form the user segmentation library;

Generating a standard word segment corresponding to the standard answer according to the standard answer, specifically including:

Segmenting the standard answer and obtaining several original standard word segments;

Filtering the plurality of original standard word segments according to each of the filter words in the filter word dictionary to obtain a plurality of the standard word segments;

A plurality of said standard word segments are combined to form the standard word segmentation library.

Further, optionally, in the method as described above, when there are multiple standard answers corresponding to the test questions, the method further includes:

The highest score is obtained from a plurality of scores that score the user answer based on each of the standard answers as the final score of the user's answer.

Further optionally, in the method as described above, according to the user segmentation database, the standard word segment corresponding to the pre-generated standard answer, and the weight of each standard word segment in the standard word segment, the user is The answer is scored, including:

All the user word segments in the user segmentation library and all standard word segments in the standard word segment library are combined and deduplicated to form a total word segmentation library;

Generating a first weight array according to a hit situation of each of the standard word segmentation in the standard word segmentation library and a weight of each of the standard word segmentation;

According to the hit situation of each user segmentation in the user segmentation library, the relationship between each user segmentation of the hit and each of the standard word segments in the standard token library, And weighting each of the standard word segments to generate a second weight array;

Calculating a cosine similarity value of the first weight array and the second weight array;

Based on the cosine similarity value, a score for scoring the user answer based on the standard answer is calculated.

The invention also provides a smart scoring device, the device comprising:

The obtaining module is configured to obtain a user answer corresponding to the question;

a generating module, configured to generate a user word segment corresponding to the user answer according to the user answer;

The scoring module is configured to score the user answer according to the user segmentation database, the standard word segment corresponding to the pre-generated standard answer, and the weight of each standard word segment in the standard word segment.

Further optionally, the device as described above further includes a setting module;

The obtaining module is further configured to obtain the test question and the standard answer corresponding to the test question;

The generating module is further configured to generate a standard word segment corresponding to the standard answer according to the standard answer;

The setting module is configured to set weights for each of the standard word segments in the standard word segmentation library.

Further optionally, the device as described above further includes a detection module;

The obtaining module is further configured to acquire an examination question used by the user to make the user answer;

The detecting module is configured to detect and determine that the test questions used by the user to make the user answer are consistent with the test questions corresponding to the standard answer.

Further optionally, in the device as described above, the setting module is specifically configured to:

Further optionally, in the device as described above, the generating module is specifically configured to:

The generating module is specifically configured to:

Further optionally, in the device as described above, the acquiring module is further configured to: when the standard answer corresponding to the question has multiple, score the user answer based on each of the standard answers The highest score is obtained from the plurality of scores as the final score of the user's answer.

Further optionally, in the apparatus as described above, the scoring module is specifically configured to:

The invention also provides a computer device, the device comprising:

One or more processors;

Memory for storing one or more programs,

The one or more programs are executed by the one or more processors such that the one or more processors implement the smart scoring method as described above.

The present invention also provides a computer readable medium having stored thereon a computer program that, when executed by a processor, implements the smart scoring method as described above.

The intelligent scoring method and device, the computer device and the computer readable medium provided by the invention obtain the user answer corresponding to the test question; generate the user word segment corresponding to the user answer according to the user answer; according to the user segmentation database, the pre-generated standard The weight of the standard word segment corresponding to the answer and the standard word segmentation in the standard segmentation database is used to score the user's answer. The technical solution of the invention can intelligently score the user's answers of the questions of the type of subjective questions, and can overcome the defects of using the method of manual marking in the prior art, resulting in a long scoring time and high labor cost. Therefore, the scoring time of the test questions can be effectively shortened, and the labor cost can be effectively saved, thereby greatly improving the scoring efficiency of the user answers of the type subjective questions.

DRAWINGS

FIG. 1 is a flowchart of Embodiment 1 of a smart scoring method according to the present invention.

FIG. 2 is a flowchart of Embodiment 2 of the smart scoring method of the present invention.

FIG. 3 is a structural diagram of Embodiment 1 of the smart scoring apparatus of the present invention.

4 is a structural diagram of a second embodiment of the smart scoring apparatus of the present invention.

Figure 5 is a block diagram of an embodiment of a computer device of the present invention.

FIG. 6 is a diagram showing an example of a computer device provided by the present invention.

detailed description

The present invention will be described in detail below with reference to the drawings and specific embodiments.

FIG. 1 is a flowchart of Embodiment 1 of a smart scoring method according to the present invention. As shown in FIG. 1 , the smart scoring method in this embodiment may specifically include the following steps:

100. Obtain a user answer corresponding to the question;

101. Generate a user word segment corresponding to the user answer according to the user answer;

102. Rate the user's answer according to the weight of the user's word segment, the standard word segment corresponding to the pre-generated standard answer, and the standard word segmentation in the standard word segment.

The execution subject of the smart scoring method in this embodiment is a smart scoring device, and the smart scoring device can be an independent electronic device for intelligently scoring the user answers of the subjective questions. The smart scoring device can be a software integrated entity to intelligently score user answers for subjective questions. The type of the test questions of this embodiment is a subjective question.

In the smart scoring method of the embodiment, when the user answer of a question is scored, the user answer corresponding to the question is obtained. For example, when the user answer of the question is in electronic form, the user answer of the question is directly read. When the user answer of the question is in the form of paper, the text recognition technology can be used to identify the text in the user's answer on the paper, and the user's answer in electronic form can be obtained. Then, according to the user's answer, the user segmentation corresponding to the user's answer can be obtained.

For example, the step 101 “generate a user word segment corresponding to the user answer according to the user answer” may specifically include the following steps:

(a1) segmentation of the user's answer, obtaining a number of original user segmentation;

(a2) filtering a plurality of original user participles according to each filter word in a preset filter word vocabulary to obtain a plurality of user participles;

(a3) Combining multiple user word segments to form a user segmentation library;

Specifically, since the user's answer may be composed of innumerable paragraphs and countless sentences, in the word segmentation, each sentence in the user's answer may be segmented. In the process of word segmentation, one sentence may be divided into one word, two words, and three. Words, four words or even multiple words; and when a word is divided, a certain participle rule can be set. For example, according to the collocation of different parts of speech in the grammar rules written in Chinese, each of the words can be reasonably The word is split to make the obtained participle more practical. The word segmentation obtained after the user's answer word segmentation is initially deduplicated to obtain the original word segmentation, and a user answer can obtain countless original word segments. Then, according to each filter word in the preset filter word lexicon, a plurality of original user word segments can be filtered to obtain a plurality of user word segments. The filter vocabulary preset in this embodiment is some words that have no practical meaning or do not contribute to the score of the user's answer. For example, the filter vocabulary can include "of", "land", "de", and "you", "me", "he", "her", "you", "us", "they", " "They" and "?", "ah", "what", "oh" and so on. According to the filter vocabulary, a plurality of user word segments can be obtained by filtering the filter words in the original user segmentation words. Finally, a plurality of user word segments are combined to form a user word segment corresponding to the user answer.

Then, according to the user word segment, the standard word segment corresponding to the pre-generated standard answer, and the weight of each standard word segment in the standard word segment, the score of the user answer based on the standard answer can be obtained. That is to say, before the user answer is scored, it is necessary to obtain the standard answer corresponding to the test question in advance, and pre-generate the standard word segment corresponding to the standard answer based on the standard answer, and configure the weight of each standard word segment in the standard word segment. Specifically, before step 102, according to the user token library, the standard word segment corresponding to the pre-generated standard answer, and the weight of each standard word segment in the standard word segment, the user answers may be scored. :

(b1) obtaining the test questions and the standard answers corresponding to the test questions;

The questions in the embodiment and the standard answers corresponding to the questions may be in electronic form or in paper form. The corresponding standard answer corresponding to the test questions and the test questions is the same as the user answer corresponding to the above test questions, and will not be described here.

(b2) generating a standard word segment corresponding to the standard answer according to the standard answer;

Correspondingly, according to the standard answer of the embodiment, the standard word segment corresponding to the standard answer is generated in the same manner as the user word segment corresponding to the user answer in the above embodiment, for example, the step (b2) is specific. It can include the following steps:

(c1) segmentation of the standard answer, obtaining several original standard word segments;

(c2) filtering a plurality of original standard word segments according to each filter word in the filter word lexicon to obtain a plurality of standard word segments;

(c3) Combine multiple standard word segments to form a standard word segmentation.

The specific implementations of the steps (c1)-(c3) are the same as the specific implementations of the foregoing steps (a1)-(a3). For details, refer to the specific implementation manners of the foregoing steps (a1)-(a3), and details are not described herein again.

(b3) Set weights for each standard word segmentation in the standard word segmentation.

In this embodiment, the weights of the standard word segmentation in the standard word segmentation library may specifically include the following two methods:

The first way: to count the frequency of occurrence of each standard participle in the standard answer, according to the frequency of occurrence of each standard participle in the standard answer, set the weight for each standard participle, so that the weight of the standard participle with high frequency is greater than the frequency of occurrence The weight of the standard participle;

For example, the frequency of each standard participle appearing in the standard answer is counted by the smart scoring device, and then each standard participle can be sorted according to the frequency of occurrence to obtain a standard word segmentation sequence; the standard participle with higher frequency can be regarded as the standard answer. The more important, key standard participles, the less frequent standard participles can be considered as non-critical standard participles that are less important to the standard answer. For example, the top N of the sorted standard word segmentation sequence may be set to a higher weight, such as a weight of 10, as a key standard word segmentation, where N may be based on the total number of standard word segments included in the standard word segmentation library. Reasonable settings, such as the number of N, can account for 30%-50% of the total number of standard word segments included in the standard word segmentation. The weights of the other standard word segments in the standard word segmentation sequence are then set to 1 as non-critical standard word segments. In this embodiment, by setting two weights as an example, in actual application, multiple file weights may be sequentially set according to the frequency of occurrence of each standard word segment in the standard answer, for example, the first 1/3 of the standard word segmentation in the standard word segmentation sequence may be used. The weight of the set is 10, the weight of the standard 1/3 of the middle 1/3 is set to 6, and the weight of the standard loyalty of the last third is 1. For the same reason, the weight of any file of 4 or more can be set. This is no longer an example.

The second way: display each standard participle for the judge to set the weight for each standard participle; and receive the weight of each standard participle input by the judger through the human interface module.

In the solution, the smart scoring device may not perform frequency statistics on the standard word segmentation in the standard word segmentation library, only display each standard word segmentation, so that the judger can view each standard word segmentation, and the human-machine interaction module can be set on the smart scoring device. For the judger to select which standard participles are the key standard participles from the standard participles displayed, which are the non-critical standard participles, and set the weights for each standard participle. Finally, the judge is returned to the smart scoring device through the human interface module of the smart scoring device. Therefore, in this embodiment, the smart scoring device may be provided with a display screen to display each standard participle. The human interface module of the smart scoring device may include a mouse and/or a keyboard for the judger to set weights for each standard word segmentation by means of a mouse and/or a keyboard. Or the human interface module of the smart scoring device may be a touch screen, and the touch screen may not only display each standard word segment, but also may have a detecting module, and input an operation for detecting the weight of the standard word segmentation of the judger on the touch screen, and receiving The input of the judger completes the setting of the weight of each standard participle. If the smart scoring device does not have a display screen, it may be provided with a communication module for transmitting to the terminal used by the judger through the communication module, and displaying the standard word segmentation on the terminal used by the judge. The judger then selects the key standard word segmentation based on the displayed standard word segmentation and the knowledge of the standard answer, and sets the weight for the key standard word segmentation. For example, the judger can select only the standard word segment with a weight of 10; Other standard word segments can be set to 10 by the smart scoring device in a pre-agreed manner. Similarly, when multiple weights need to be set for multiple standard word segments, the judge can also select the most critical one, two or more standard word segments, and then set the weight of the selected standard word segmentation to 10; Then select the second key class of standard word segmentation, the corresponding can be set to a weight less than 10, and so on, according to the reduction of the key degree of each standard word segmentation, the value of the weight can be set smaller in turn. And in the end, there are some standard word segments that are not selected by the user, which can be regarded as unimportant standard word segments, and a smaller weight can be uniformly set, for example, 1.

In addition, optionally, the first mode and the second mode may also be combined, and the smart scoring device first sets the weight of each standard word segment according to the first manner, and then displays the standard word segmentation according to the second manner. The weight is used for the judge to check whether the weight of each standard participle set by the smart scoring device is reasonable. If it is unreasonable, the judge can modify the weight of the standard word segment according to the human interface module. For example, if a standard participle should be a keyword in the standard answer, the weight should be set to the maximum weight, such as 10, but the smart scoring device can only set its weight to 6 based on the result of the frequency statistics. At this time, the judger The weight of the standard word segmentation can be changed from 6 to 10 according to the weight of the displayed standard word segmentation.

According to the manner of the foregoing embodiment, the user word segment corresponding to the user answer of the question, the standard word segment corresponding to the standard answer of the test question, and the weight of each standard word segment in the standard word segment can be obtained, because the user answer of most users It is not the same as the standard answer, so the user segmentation form formed according to the user's answer is the same as or different from the participle included in the standard word segment formed by the standard answer; in the present embodiment, based on the When the standard answer is the score of the user's answer, the weight of each standard word segment is also considered, so that if the user participle in the user's word segment hits the standard participle of the weight in the standard word segment, a higher score can be obtained. The user participle in the thesaurus only hits the standard participle with a small weight in the standard word segmentation library, and can obtain a lower score, so that the score made by the smart scoring device is more reasonable, and the obtained score is more accurate.

The intelligent scoring method of the embodiment obtains the user answer corresponding to the test question; generates a user word segment corresponding to the user answer according to the user answer; and the standard word segment and the standard score corresponding to the user word segment library, the pre-generated standard answer; The weight of each standard participle in the thesaurus is scored for the user's answer. By adopting the technical solution of the embodiment, the user answers of the questions of the type of subjective questions can be intelligently scored intelligently, which can overcome the method of manual marking in the prior art, resulting in a longer scoring time and higher labor cost. The defects can effectively shorten the scoring time of the questions, and can also effectively save labor costs, thereby greatly improving the scoring efficiency of the user answers to the type of subjective questions.

FIG. 2 is a flowchart of Embodiment 2 of the smart scoring method of the present invention. As shown in FIG. 2, the smart scoring method of the present embodiment is based on the technical solution of the foregoing embodiment, and the technical solution of the present invention is described by taking an example that the test questions include only one standard answer. As shown in FIG. 2, the smart scoring method of this embodiment may specifically include the following steps:

200. Acquiring the type of questions for the subjective question and the standard answer corresponding to the question;

The execution subject of the smart scoring method of this embodiment is still a smart scoring device. For the method of the present embodiment and the method for obtaining the standard answer corresponding to the question, reference may be made to the manner of obtaining the user answer and the standard answer in the above embodiment, and details are not described herein again.

201. Generate a standard word segment corresponding to the standard answer according to the standard answer;

In this embodiment, taking only one standard answer as an example, after obtaining the standard answer, the standard word segment corresponding to the standard answer may be obtained according to steps (c1)-(c3) in the above embodiment.

202. Set weights for each standard word segmentation in the standard word segmentation library;

For the manner of setting the weights, refer to the two methods of setting weights in the foregoing embodiments, and details are not described herein again.

203. Obtain an examination question used by the user to make a user answer;

Similarly, the questions used by the user to make the user's answer can also be in electronic form or in paper form.

204. Detect whether the test questions used by the user to make the user's answer are consistent with the test questions corresponding to the obtained standard answer; if they are consistent, perform step 205; otherwise, determine that the test questions are inconsistent and exit the current round of the score.

205. Obtain a user answer corresponding to the test question;

206. Generate a user word segment corresponding to the user answer according to the user answer;

In this embodiment, the process of obtaining the user's answer is the same as that of the above embodiment. After the user's answer is obtained, the user word segment corresponding to the user answer may be generated according to steps (a1)-(a3) in the above embodiment.

207. Combine and de-weight all user participles in the user segmentation library and all standard word segments in the standard word segmentation library to form a total word segmentation library;

208. Generate a first weight array according to a hit of the standard word segmentation in the standard token segment and a weight of each standard word segment;

209. Generate a second according to the hit situation of each user segmentation in the user segmentation database, the relationship between each user segmentation of the hit and the standard word segmentation in the standard token segment, and the weight of each standard segmentation word. Weighted array

210. Calculate a cosine similarity value of the first weight array and the second weight array;

211. Calculate a score for scoring the user's answer based on the standard answer according to the cosine similarity value.

Steps 207-211 of the embodiment are the steps 102 of the foregoing embodiment shown in FIG. 1 "weighting according to the user segmentation database, the standard word segment corresponding to the pre-generated standard answer, and the standard word segmentation in the standard word segment. , a specific implementation of scoring user answers.

For the same question, the user's user answer cannot be completely deviated from the standard answer. Therefore, there are always some user word segmentation in the user segmentation database corresponding to the user's answer that can hit the standard word segmentation in the standard word segmentation library, that is, the user word segmentation library. Some of the user participles in the paper will coincide with some of the standard participles in the standard participle. After all the user word segments in the user segmentation are combined with all the standard word segments in the standard segmentation library, they are deduplicated to form a total word segmentation. The deduplication here means that if the user participle is the same as the standard participle, only one of them is retained, and the participle in the total distractor obtained is not repeated. Then, the standard word segment is compared with the total word segment, and the first weight array is generated according to the hits of the standard word segments in the standard word segment in the total word segment and the weights of the standard word segments; The number of elements in the weighted array is the same as the number of participles included in the total token bank. For example, if the standard participle M in a standard word segment hits the participle M in the total token bank, the standard participle M has a corresponding weight of 10 in the standard token library, and the position corresponding to the participle M in the first weight array Above, the set weight value is the weight of the standard participle M. For the participles in the total token bank that are not hit by the standard participle in the standard token library, the weight set in the corresponding position in the first weight array is 0. That is, if the standard participle in the standard participle hits the participle in the total participle, in the first weight array, the hit participle corresponds to the position where the hit standard participle is set in advance; and the total participle The weights are set to 0 at the positions corresponding to the remaining missed participles in the library.

In the same way, according to the user's word segmentation in the user segmentation database in the total segmentation database, the hit user segmentation and the standard word segmentation in the standard word segmentation, and the weight of each standard word segmentation , generating a second weight array. Similarly, the number of elements in the second weight array is the same as the number of participles included in the total token library. When the user segmentation library is compared with the total token segment, if the user segmentation N in a user segmentation hits the segmentation word N in the total token segment, and the hit user segmentation is the standard segmentation in the standard segmentation segment. The weight corresponding to the standard participle can be obtained. At this time, in the second weight array, the weight of the corresponding standard word segment is set at the position corresponding to the hit participle. If the user participle N in a user participle hits the participle N in the total tokenizer, but the hit user participle is not the standard participle in the standard token pool, there is no pre-set weight, and the second weight is In the array, the weighted 1 is set at the position corresponding to the hit participle. Finally, the weight is set to 0 for the position corresponding to the remaining missed participles in the total token bank.

For the convenience of description, the standard participle corresponding to the standard answer below includes four standard parts of A, B, C and D, wherein the key standard participles A and B are set to a weight of 10, and the non-critical standard participles C and D. The weight of the user is 1 as an example. The user segmentation corresponding to the user answer includes four user segments A, B, E, and F as an example. In this case, the total segmentation database includes A, B, C, D, and E. There are 6 parts of the word segmentation with F; it can be known that the first weight array and the second weight array contain 6 elements. Compare A, B, C and D in the standard word segment with A, B, C, D, E and F in the total word segment, where the standard participle A in the standard participle hits the total participle The participle A in the library, at this time, obtains the weight 10 corresponding to the standard participle A, and sets the weight of the position corresponding to the participle A in the first weight array to 10. In the same way, the position corresponding to the participle B in the first weight array can be set to a weight of 10, and the position corresponding to the participles C and D can be set to a weight of 1. The participles E and F in the total word segment are not hit by the standard participle in the standard participle. The weights of the parts E and F in the first weight array can be set to 0, so that the first weight array is [10,10,1,1,0,0].

Compare A, B, E, and F in the user segmentation database with A, B, C, D, E, and F in the total segmentation database, where the user segmentation A in the user segmentation library hits the total segmentation The participle A in the library, and the user participle A is also a standard participle in the standard token library, and the weight 10 of the standard participle A is obtained, and the position corresponding to the participle A in the second weight array is set to be 10. In the same way, the position weight corresponding to the participle B in the second weight array can also be set to 10; the user participles E and F also hit the participles E and F in the total word segment respectively, but the user participles A and F are not standard. The word segmentation in the word segment, in which case the weights of the parts E and F in the second weight array can be set to 1; and the word parts C and D in the total word segment are not user participles in the user word segment. Hit, you can set the weight to 0 in the position corresponding to the participles C and D in the second weight array, so that the first weight array obtained is [10,10,0,0,1,1].

In the above manner, the first weight array and the second weight array can be obtained, and then the cosine similarity values of the first weight array and the second weight array are calculated; for example, the specific formula can be calculated by:

Where S is the first weight array, which represents the weight information of the hits of the standard participle in the standard token library in the total token bank; n is the number of elements in the first weight array. The first weight array may be represented as [S ₁ , S ₂ , ..., S _n ]; where S _i represents the ith element in the first weight array.

U is a second weight array, which represents the weight information of the hits of the user participle in the user token pool in the total token bank; n is the number of elements in the second weight array. The second weighted array can be represented as [U ₁ , U ₂ , . . . , U _n ]; where U _i represents the ith element in the second weighted array.

After the calculation of the above formula, the cosine similarity value cos θ of the first weight array and the second weight array can be obtained, that is, the corresponding result is a value between 0 and 1; the obtained cosine similarity value is closer to 1, indicating The closer the user's answer is to the standard answer. If the result is 1, it means that the user's answer can be approximated to the standard answer, and the question can get full marks. Then, based on the cosine similarity value, a score for scoring the user's answer based on the standard answer is calculated. For example, the cosine similarity value may be multiplied by the total score of the question as a score for scoring the user's answer based on the standard answer. For example, if the score of the question is 100 points and the obtained cosine similarity value is 0.81, the score of the user's answer may be 0.81×100=81 points.

In the above embodiment, the whole process of scoring a user's answer is included by including only one standard answer; in the above manner, all user answers can be scored in turn.

In addition, a plurality of standard answers may be included in the same question. Based on each standard answer, the user answers may be scored according to the manner of the above embodiment to obtain a score. For the same user answer, according to the above embodiment, the scores can be scored based on a plurality of standard answers, respectively, to obtain a plurality of scores. Finally, the highest score is obtained from the scores that score the user's answer based on the standard answers, as the final score of the user's answer.

The intelligent scoring method of the embodiment can intelligently score the user answers of the questions of the type subjective questions by using the above technical solution, which can overcome the prior art method of using manual scoring, resulting in a higher scoring time. Long, high labor cost defects, which can effectively shorten the scoring time of the questions, can also effectively save labor costs, and thus greatly improve the scoring efficiency of the user answers to the type of subjective questions.

FIG. 3 is a structural diagram of Embodiment 1 of the smart scoring apparatus of the present invention. As shown in FIG. 3, the smart scoring apparatus of this embodiment may specifically include: an obtaining module 10, a generating module 11, and a scoring module 12.

The obtaining module 10 is configured to obtain a user answer corresponding to the question; the generating module 11 is configured to generate a user word segment corresponding to the user answer according to the user answer obtained by the obtaining module 10; the scoring module 12 is configured to generate the user segment according to the generating module 11 The user's answer is scored by the lexicon, the standard word segment corresponding to the pre-generated standard answer, and the weight of each standard word segment in the standard word segment.

The smart scoring apparatus of the present embodiment is the same as the implementation of the foregoing method embodiment by using the above-mentioned modules to implement the smart scoring principle. For details, refer to the description of the related method embodiments, and details are not described herein again.

4 is a structural diagram of a second embodiment of the smart scoring apparatus of the present invention. As shown in FIG. 4, the smart scoring apparatus of the present embodiment further includes the following technical solutions based on the technical solution of the embodiment shown in FIG.

As shown in FIG. 4, the smart scoring apparatus of this embodiment further includes a setting module 13.

The obtaining module 10 is further configured to obtain a standard answer corresponding to the test question and the test question; the generating module 11 is further configured to generate a standard word segment corresponding to the standard answer according to the standard answer obtained by the obtaining module 10; and the setting module 13 is configured to generate the module 11 The standard word segmentation in the generated standard word segment sets weights.

Further optionally, as shown in FIG. 4, the smart scoring device of the embodiment further includes a detecting module 14.

The obtaining module 10 is further configured to obtain a question used by the user to make a user answer; the detecting module 14 is configured to detect and determine that the question used by the user of the obtaining module 10 to make the user answer is consistent with the question corresponding to the standard answer.

Further, in the smart scoring device of the embodiment, the setting module 13 is specifically configured to:

Count the frequency of occurrence of each standard participle in the standard answer, and set the weight for each standard participle according to the frequency of occurrence of each standard participle in the standard answer, so that the weight of the standard participle with high frequency is greater than the weight of the standard participle with low frequency;

Or display each standard participle for the judge to set the weight for each standard participle; and receive the weight of each standard participle input by the judger through the human interface module.

Further, optionally, in the smart scoring device of the embodiment, the generating module 11 is specifically configured to:

Segmentation of user answers, obtaining several original user word segments;

Filtering a plurality of original user word segments according to each filter word in a preset filter word vocabulary to obtain a plurality of user word segments;

Combine multiple user participles to form a user segmentation library;

Further, optionally, in the smart scoring apparatus of the embodiment, the generating module 11 is further configured to:

Segment the standard answer and obtain several original standard word segments;

Filtering a plurality of original standard word segments according to each filter word in the filter word lexicon to obtain a plurality of standard word segments;

Combine multiple standard word segments to form a standard word segmentation.

Further optionally, in the smart scoring device of the embodiment, the obtaining module 10 is further configured to obtain the highest score from the plurality of scores that are scored for the user answer based on the standard answers when there are multiple standard answers corresponding to the question. Value, as the final score for the user's answer.

Further, in the smart scoring device of the embodiment, the scoring module 12 is specifically configured to:

All user participles in the user segmentation library and all standard word segments in the standard word segmentation library are combined and deduplicated to form a total word segmentation library;

Generating a first weight array according to the hits of the standard word segments in the standard token segment in the total token bank and the weights of the standard word segments;

Generating a second weight array according to the hit situation of each user participle in the user segmentation database in the total token segment, the relationship between each user segmentation of the hit and the standard word segmentation in the standard token library, and the weight of each standard word segmentation. ;

A score that scores the user's answer based on the standard answer is calculated based on the cosine similarity value.

Figure 5 is a block diagram of an embodiment of a computer device of the present invention. As shown in FIG. 5, the computer device of the present embodiment includes: one or more processors 20, and a memory 30 for storing one or more programs when one or more programs stored in the memory 30 are The one or more processors 20 execute such that the one or more processors 20 implement the smart scoring method of the embodiment illustrated in Figures 1-2 above. In the embodiment shown in FIG. 5, a plurality of processors 20 are included as an example.

For example, FIG. 6 is a diagram showing an example of a computer device provided by the present invention. FIG. 6 shows a block diagram of an exemplary computer device 12a suitable for use in implementing embodiments of the present invention. The computer device 12a shown in FIG. 6 is merely an example and should not impose any limitation on the function and scope of use of the embodiments of the present invention.

As shown in Figure 6, computer device 12a is embodied in the form of a general purpose computing device. The components of computer device 12a may include, but are not limited to, one or more processors 16a, system memory 28a, and bus 18a that connects different system components, including system memory 28a and processor 16a.

Bus 18a represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of a variety of bus structures. For example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA Bus, a Video Electronics Standards Association (VESA) local bus, and peripheral component interconnects ( PCI) bus.

Computer device 12a typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computer device 12a, including volatile and nonvolatile media, removable and non-removable media.

System memory 28a may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30a and/or cache memory 32a. Computer device 12a may further include other removable/non-removable, volatile/non-volatile computer system storage media. By way of example only, storage system 34a may be used to read and write non-removable, non-volatile magnetic media (not shown in Figure 6, commonly referred to as "hard disk drives"). Although not shown in FIG. 6, a disk drive for reading and writing to a removable non-volatile disk (such as a "floppy disk"), and a removable non-volatile disk (such as a CD-ROM, DVD-ROM) may be provided. Or other optical media) read and write optical drive. In these cases, each drive can be coupled to bus 18a via one or more data medium interfaces. System memory 28a may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the various embodiments of Figures 1-4 of the present invention described above.

A program/utility 40a having a set (at least one) of program modules 42a, which may be stored, for example, in system memory 28a, including, but not limited to, an operating system, one or more applications, others Program modules and program data, each of these examples or some combination may include an implementation of a network environment. Program module 42a typically performs the functions and/or methods of the various embodiments of Figures 1-4 described above with respect to the present invention.

Computer device 12a may also be in communication with one or more external devices 14a (eg, a keyboard, pointing device, display 24a, etc.), and may also be in communication with one or more devices that enable a user to interact with the computer device 12a, and/or Any device (eg, a network card, modem, etc.) that enables the computer device 12a to communicate with one or more other computing devices. This communication can take place via an input/output (I/O) interface 22a. Also, computer device 12a may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and/or a public network, such as the Internet) through network adapter 20a. As shown, network adapter 20a communicates with other modules of computer device 12a via bus 18a. It should be understood that although not shown in the figures, other hardware and/or software modules may be utilized in conjunction with computer device 12a, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives. And data backup storage systems, etc.

The processor 16a executes various function applications and data processing by running a program stored in the system memory 28a, for example, implementing the obstacle recognition method shown in the above embodiment.

The present invention also provides a computer readable medium having stored thereon a computer program that, when executed by a processor, implements the smart scoring method as shown in the above embodiments.

The computer readable medium of the present embodiment may include the RAM 30a, and/or the cache memory 32a, and/or the storage system 34a in the system memory 28a in the embodiment shown in FIG. 6 described above.

With the development of technology, the propagation of computer programs is no longer limited by tangible media, and can be downloaded directly from the network or obtained in other ways. Therefore, the computer readable medium in this embodiment may include not only a tangible medium but also an intangible medium.

The computer readable medium of this embodiment can employ any combination of one or more computer readable mediums. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive lists) of computer readable storage media include: electrical connections having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read only memory (ROM), Erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing. In this document, a computer readable storage medium can be any tangible medium that can contain or store a program, which can be used by or in connection with an instruction execution system, apparatus or device.

A computer readable signal medium may include a data signal that is propagated in the baseband or as part of a carrier, carrying computer readable program code. Such propagated data signals can take a variety of forms including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the foregoing. The computer readable signal medium can also be any computer readable medium other than a computer readable storage medium, which can transmit, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device. .

Program code embodied on a computer readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including an object oriented programming language such as Java, Smalltalk, C++, and conventional A procedural programming language - such as the "C" language or a similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer, partly on the remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (eg, using an Internet service provider) Internet connection).

In the several embodiments provided by the present invention, it should be understood that the disclosed system, apparatus, and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division, and the actual implementation may have another division manner.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

The above-described integrated unit implemented in the form of a software functional unit can be stored in a computer readable storage medium. The above software functional unit is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to perform the methods of the various embodiments of the present invention. Part of the steps. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes. .

The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalents, improvements, etc., which are made within the spirit and principles of the present invention, should be included in the present invention. Within the scope of protection.

Claims

An intelligent scoring method, the method comprising:

Obtain the user answer corresponding to the question;

Generating a user word segment corresponding to the user answer according to the user answer;

The user answers are scored according to the user segmentation database, the standard word segment corresponding to the pre-generated standard answer, and the weights of the standard word segments in the standard word segment.
The method according to claim 1, wherein the basis is obtained according to the weight of the user segmentation library, the standard word segment corresponding to the pre-generated standard answer, and the standard word segmentation in the standard word segment library. Before the standard answer is a score for scoring the user answer, the method further includes:

Obtaining the test question and the standard answer corresponding to the test question;

Generating a standard word segment corresponding to the standard answer according to the standard answer;

A weight is set for each of the standard word segments in the standard word segmentation.
The method according to claim 2, wherein the method further comprises: before generating the user token library corresponding to the user answer according to the user answer, the method further comprising:

Obtaining the questions used by the user to make the user's answer;

Detecting and determining that the test questions used by the user to make the user answer are consistent with the test questions corresponding to the standard answer.
The method according to claim 2 or 3, wherein the weighting is set for each of the standard participles in the standard word segment, specifically comprising:

Counting the frequency of occurrence of each of the standard participles in the standard answer, and setting weights for each of the standard participles according to the frequency of occurrence of each of the standard participles in the standard answer, so that the standard participle with high frequency of occurrence occurs The weight of the standard is greater than the weight of the standard participle with a low frequency of occurrence;

Or displaying each of the standard participles for the judge to set a weight for each of the standard participles; and receiving the weight of each of the standard participles input by the judger through the human interface module.
The method according to claim 2, 3 or 4, wherein the generating a user word segment corresponding to the user answer according to the user answer comprises:

Segmenting the user's answer and obtaining a number of original user word segments;

Filtering the plurality of original user participles according to each filter word in the preset filter word lexicon to obtain a plurality of the user participles;

Combining a plurality of the user participles to form the user segmentation library;

Generating a standard word segment corresponding to the standard answer according to the standard answer, specifically including:

Segmenting the standard answer and obtaining several original standard word segments;

Filtering the plurality of original standard word segments according to each of the filter words in the filter word dictionary to obtain a plurality of the standard word segments;

A plurality of said standard word segments are combined to form the standard word segmentation library.
The method according to any one of claims 1 to 5, wherein when there are a plurality of the standard answers corresponding to the questions, the method further comprises:

The highest score is obtained from a plurality of scores that score the user answer based on each of the standard answers as the final score of the user's answer.
The method according to any one of claims 1 to 6, wherein the standard word segment corresponding to the user word segment, the pre-generated standard answer, and the standard word segmentation in the standard word segment are Weights, which are scored for the user's answers, including:

All the user word segments in the user segmentation library and all standard word segments in the standard word segment library are combined and deduplicated to form a total word segmentation library;

Generating a first weight array according to a hit situation of each of the standard word segmentation in the standard word segmentation library and a weight of each of the standard word segmentation;

According to the hit situation of each user segmentation in the user segmentation library, the relationship between each user segmentation of the hit and each of the standard word segments in the standard token library, And weighting each of the standard word segments to generate a second weight array;

Calculating a cosine similarity value of the first weight array and the second weight array;

Based on the cosine similarity value, a score for scoring the user answer based on the standard answer is calculated.
A smart scoring device, the device comprising:

The obtaining module is configured to obtain a user answer corresponding to the question;

a generating module, configured to generate a user word segment corresponding to the user answer according to the user answer;

The scoring module is configured to score the user answer according to the user segmentation database, the standard word segment corresponding to the pre-generated standard answer, and the weight of each standard word segment in the standard word segment.
The device according to claim 8, wherein said device further comprises a setting module;

The obtaining module is further configured to obtain the test question and the standard answer corresponding to the test question;

The generating module is further configured to generate a standard word segment corresponding to the standard answer according to the standard answer;

The setting module is configured to set weights for each of the standard word segments in the standard word segmentation library.
The device according to claim 9, wherein said device further comprises a detection module;

The obtaining module is further configured to acquire an examination question used by the user to make the user answer;

The detecting module is configured to detect and determine that the test questions used by the user to make the user answer are consistent with the test questions corresponding to the standard answer.
The device according to claim 9 or 10, wherein the setting module is specifically configured to:

Counting the frequency of occurrence of each of the standard participles in the standard answer, and setting weights for each of the standard participles according to the frequency of occurrence of each of the standard participles in the standard answer, so that the standard participle with high frequency of occurrence occurs The weight of the standard is greater than the weight of the standard participle with a low frequency of occurrence;

Or displaying each of the standard participles for the judge to set a weight for each of the standard participles; and receiving the weight of each of the standard participles input by the judger through the human interface module.
The device according to claim 9, 10 or 11, wherein the generating module is specifically configured to:

Segmenting the user's answer and obtaining a number of original user word segments;

Filtering the plurality of original user participles according to each filter word in the preset filter word lexicon to obtain a plurality of the user participles;

Combining a plurality of the user participles to form the user segmentation library;

The generating module is specifically configured to:

Segmenting the standard answer and obtaining several original standard word segments;

And filtering the plurality of original standard word segments according to each of the filter words in the filter word dictionary to obtain a plurality of the standard word segments;

A plurality of said standard word segments are combined to form the standard word segmentation library.
The device according to any one of claims 8 to 12, wherein the obtaining module is further configured to: when there are multiple standard answers corresponding to the question, The highest score is obtained from the plurality of scores in which the user answers are scored as the final score of the user's answer.
The apparatus according to any one of claims 8 to 13, wherein the scoring module is specifically configured to:

All the user word segments in the user segmentation library and all standard word segments in the standard word segment library are combined and deduplicated to form a total word segmentation library;

Generating a first weight array according to a hit situation of each of the standard word segmentation in the standard word segmentation library and a weight of each of the standard word segmentation;

According to the hit situation of each user segmentation in the user segmentation library, the relationship between each user segmentation of the hit and each of the standard word segments in the standard token library, And weighting each of the standard word segments to generate a second weight array;

Calculating a cosine similarity value of the first weight array and the second weight array;

Based on the cosine similarity value, a score for scoring the user answer based on the standard answer is calculated.
A computer device, characterized in that the device comprises:

One or more processors;

Memory for storing one or more programs,

The one or more programs are executed by the one or more processors such that the one or more processors implement the method of any of claims 1-7.
A computer readable medium having stored thereon a computer program, characterized in that the program, when executed by a processor, implements the method of any of claims 1-7.