TW201443667A - Artificial intelligent test paper item system and item method thereof - Google Patents
Artificial intelligent test paper item system and item method thereof Download PDFInfo
- Publication number
- TW201443667A TW201443667A TW102116514A TW102116514A TW201443667A TW 201443667 A TW201443667 A TW 201443667A TW 102116514 A TW102116514 A TW 102116514A TW 102116514 A TW102116514 A TW 102116514A TW 201443667 A TW201443667 A TW 201443667A
- Authority
- TW
- Taiwan
- Prior art keywords
- test
- questions
- difficulty
- module
- degree
- Prior art date
Links
Landscapes
- Electrically Operated Instructional Devices (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
本發明係一種試卷組題系統,尤指一種利用人工智慧依照所選鑑別度分類考題難易度的試卷組題系統。 The invention relates to a test paper group question system, in particular to a test paper group question system which uses artificial intelligence to classify test questions according to the selected degree of discrimination.
現有學校或教育單位為確認學生是否了解老師的授課內容,以及令老師了解學生於授課後的學習成效,因此於各學期皆設有多次的考試評量,並依照評量結果針對不同學生的學習程度進行輔導,由於各個學生的程度與學習成效不同,因此老師欲從考試題庫中挑選適合的考試題目,以作為學生學習鑑別度時,需要針對學生程度與考試題目花費許多時間進行篩選分類,其中尤以國中與國小的題庫最為龐大且繁雜,欲組合或挑選出適合班上各個學生程度的考卷實不容易。 In order to confirm whether the student understands the teacher's lecture content and to let the teacher understand the student's learning outcome after the lecture, the existing school or educational unit has multiple test evaluations in each semester, and according to the evaluation results for different students. The degree of learning is used for counseling. Because the degree of each student is different from the learning outcome, the teacher wants to select suitable test questions from the test question bank. As a student learning degree, it takes a lot of time to filter and classify the student degree and the test title. Among them, the question bank of the middle and the small country is the most large and complicated, and it is not easy to combine or select the examination papers suitable for each student level in the class.
由於受限於授課老師個人的專業素養、過往教學或出題的經驗,容易主觀地判斷該次考試出題的難度,因此造成不同出題老師選出之題目的難易度差別過大,而有鑑別度不佳的問題,以及學生考試結果落差過大造成考試成績不具有參考價值的缺點。 Due to the limited professionalism, past teaching or experience of the lecturer, it is easy to subjectively judge the difficulty of the test, so the difficulty of the questions selected by different teachers is too large, and the discrimination is not good. The problem, as well as the gap between the student's test results and the test results are not worthy of reference.
如前揭所述,現有考題容易受出題人員之專業素養及個人主觀因素影響,產生考題難易度差別過大,而有鑑別度不佳的問題,因此本發明主要目的在提供一人工智慧試卷組題系統及其組題方法,主要是利用人工智慧分析試題難度,依據所選鑑別度由題庫中挑選對應難度的考題並組成試卷,解決現有試卷由人工分類篩選產生之鑑別度不佳的問題。 As mentioned above, the existing examination questions are easily affected by the professional literacy and personal subjective factors of the students, and the differences in the difficulty of the examination questions are too large, and there are problems of poor discrimination. Therefore, the main purpose of the present invention is to provide an artificial wisdom test paper group. The system and its group method are mainly used to analyze the difficulty of the test questions by artificial intelligence. According to the selected degree of discrimination, the questions of the corresponding difficulty are selected from the question bank and the test papers are composed to solve the problem that the existing test papers are poorly discriminated by manual classification and screening.
為達成前述目的所採取的主要技術手段係令前述人工智慧試卷組題系統,包含有:一資料庫,其儲存有多數個試題;一題目難易鑑別度模組,其與資料庫連結,用以設定各試題的難易度及所屬範圍;一題目篩選模組,其與資料庫連結,用以根據所選鑑別度挑選資料庫中對應難度的試題,並組成具多數考題的試卷;一考試結果驗證模組,其與資料庫連結,用以取得外部一使用者對各考題的答案;以及一考生程度判別模組,其分別與題目難易鑑別度模組以及考試結果驗證模組連結,用以依據試題難度與考題答案判斷學生程度。 The main technical means adopted to achieve the foregoing objectives is that the aforementioned artificial intelligence test paper group title system includes: a database storing a plurality of test questions; a difficult-to-discriminate degree module connected to the database for Set the difficulty level and scope of each test question; a topic screening module, which is linked with the database, is used to select the test questions corresponding to the difficulty in the database according to the selected degree of discrimination, and form a test paper with a majority of questions; The module is connected to the database for obtaining an external user's answer to each question; and a candidate degree discriminating module, which is respectively connected with the problem difficulty identification module and the test result verification module, for The difficulty of the test questions and the answers to the questions determine the degree of students.
為達成前述目的所採取的主要技術手段係令前述人工智慧試卷組題方法,包含有: 設定試卷的鑑別度;判斷題庫內試題數量是否滿足考試之鑑別度所需試題數量,若題庫內試題數量足夠,則依所設之鑑別度直接組成考題並進行人工智慧驗證訓練,若題庫內試題數量不足,則隨機出題並進行難易鑑別度訓練;以及依照組成考題、答題成績與考生程度進行人工智慧訓練,使下次試卷的組題難度與鑑別度符合考生程度。 The main technical means adopted to achieve the above objectives is the method of the aforementioned artificial wisdom test paper group, which includes: Set the discriminating degree of the test paper; determine whether the number of questions in the question bank meets the number of questions required for the discriminating degree of the test. If the number of questions in the question bank is sufficient, directly form the test questions and perform artificial intelligence verification training according to the set discrimination degree, if the test questions in the question bank If the number is insufficient, the questions will be randomly selected and the difficulty discrimination training will be carried out; and the artificial intelligence training will be carried out according to the test questions, the results of the questions and the degree of the candidates, so that the difficulty and discrimination of the group questions of the next test paper are in accordance with the degree of candidates.
藉由前述元件組成的人工智慧試卷組題系統,利用人工智慧分析題庫中之各試題的難度,並依據所選鑑別度的高低,由具有大量試題之題庫中挑選對應難度的試題並組成試卷,可達到最高鑑別度,使所有考生的分數呈現常態分布,而為一份有效區分各考生程度的試卷,測驗完畢後分析成績與考生程度並進行人工智慧訓練,使下次考試可準確地依照需求或所選鑑別度給與適當難度題目,並降低老師篩選考題時間,解決現有試卷由人工分類產生之鑑別度不佳的問題。 The artificial intelligence test paper group question system consisting of the aforementioned components uses artificial intelligence to analyze the difficulty of each test question in the question bank, and according to the selected degree of discrimination, selects the test questions corresponding to the difficulty from the question bank with a large number of test questions and composes the test paper. The highest degree of discrimination can be achieved, so that all candidates' scores are normally distributed, and a test paper that effectively distinguishes the degree of each candidate is analyzed. After the test is completed, the scores and candidates are analyzed and artificial intelligence training is performed, so that the next test can be accurately determined according to the needs. Or the selected degree of discrimination is given to the appropriate difficulty topic, and the teacher's screening test time is reduced, and the problem of poor discrimination of the existing test paper by manual classification is solved.
《本發明》 "this invention"
10‧‧‧伺服器 10‧‧‧Server
10A‧‧‧電子裝置 10A‧‧‧Electronic devices
11‧‧‧操作介面 11‧‧‧Operator interface
20‧‧‧資料庫 20‧‧‧Database
30‧‧‧題目難易鑑別度模組 30‧‧‧Difficult identification module
40‧‧‧題目篩選模組 40‧‧‧Title screening module
50‧‧‧考試結果驗證模組 50‧‧‧Exam results verification module
60‧‧‧考生程度判別模組 60‧‧‧ Candidate Level Discrimination Module
圖1是本發明較佳實施例的電路方塊圖。 BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a block diagram of a circuit in accordance with a preferred embodiment of the present invention.
圖2是本發明較佳實施例的流程圖。 2 is a flow chart of a preferred embodiment of the present invention.
圖3是本發明較佳實施例之電子裝置的操作畫面示意圖。 3 is a schematic diagram of an operation screen of an electronic device according to a preferred embodiment of the present invention.
圖4是本發明較佳實施例之成績分佈示意圖。 4 is a schematic diagram showing the distribution of achievements in a preferred embodiment of the present invention.
圖5是本發明較佳實施例之難度與鑑別度分佈示意圖。 Figure 5 is a schematic diagram showing the distribution of difficulty and discrimination in a preferred embodiment of the present invention.
關於本發明的較佳實施例,請參閱圖1至3所示,係於一伺服器10中設有一資料庫20、一題目難易鑑別度模組30、一題目篩選模組40、一考試結果驗證模組50以及一考生程度判別模組60,於本較佳實施例中,該伺服器10連結有一個以上的電子裝置10A,該電子裝置10A係透過網路連線存取伺服器10的資料,以顯示操作畫面、試卷或考題內容,該電子裝置10A係一智慧型手機、一平板電腦、一筆記型電腦或一桌上型電腦。 For a preferred embodiment of the present invention, please refer to FIG. 1 to FIG. 3, which is provided with a database 20, a difficult identification module 30, a title screening module 40, and an examination result in a server 10. In the preferred embodiment, the server 10 is connected to one or more electronic devices 10A, and the electronic device 10A accesses the server 10 through a network connection. The electronic device 10A is a smart phone, a tablet computer, a notebook computer or a desktop computer, for displaying an operation screen, a test paper or a test item.
該資料庫20係儲存有多數個測驗試題,並將該等試題分為不同年級與科目,例如:一年級、二年級或六年級,以及自然科、國文科或數學科。 The database 20 stores a plurality of quiz questions and classifies the questions into different grades and subjects, such as first grade, second grade or sixth grade, and natural sciences, national liberal arts or mathematics.
該題目難易鑑別度模組30係與資料庫20連結,其設定資料庫20中之各個試題的難易度,例如:難、中或易。 The problem difficulty identification module 30 is linked to the database 20, which sets the difficulty of each test question in the database 20, for example, difficult, medium or easy.
該題目篩選模組40係與資料庫20連結,用以根據所選鑑別度挑選資料庫20中對應難度的試題,並組成具多數考題的試卷。於本較佳實施例中,該題目篩選模組40係一具有人工智慧的支持向量機(Support Vector Machine),對考題進行機器學習訓練,可選定各種訓練目標如期望值、中位數、眾數、變異數、偏度、峰度,當一 份已訓練完整的題庫,可依照老師意願,篩選出班級內的資優生、落後學生、或單純以高鑑別度方式選出考題以檢視全班能力。其中,鑑別度調整係指定考題鑑別度,使考試成績適用於鑑別資優生、落後學生或一般常態分布,單一學生訓練係依照單一學生狀況,組合出適合學生程度的考題並設定難度,使學生考試成績接近設定的分數,考試結果並可回送支持向量機,將結果繼續訓練。 The title screening module 40 is coupled to the database 20 for selecting test questions of the corresponding difficulty in the database 20 according to the selected degree of discrimination, and forming a test paper with a majority of questions. In the preferred embodiment, the title screening module 40 is a support vector machine with artificial intelligence, and performs machine learning training on the questions, and can select various training targets such as expected value, median, and mode. , variation, skewness, kurtosis, when A well-trained test bank can be selected according to the teacher's wishes to select the gifted students, backward students, or simply select questions in a high-discrimination manner to examine the class's ability. Among them, the degree of discrimination adjustment specifies the degree of discrimination of the test questions, so that the test scores are suitable for identifying gifted students, backward students or general normal distribution. The single student training system combines the questions suitable for the students according to the single student situation and sets the difficulty to make the students test. The score is close to the set score, the test results can be returned to the support vector machine, and the results will continue to be trained.
該考試結果驗證模組50係與資料庫20連結,用以取得外部一使用者對各考題的答案。 The test result verification module 50 is linked to the database 20 for obtaining an external user's answer to each question.
該考生程度判別模組60係分別與題目難易鑑別度模組30以及考試結果驗證模組50連結,用以依據題目難度與答案判斷學生程度。 The candidate degree discriminating module 60 is respectively connected to the topic difficulty identification module 30 and the test result verification module 50 for judging the student degree according to the difficulty and the answer of the question.
如圖2、3所示,使用者透過電子裝置10A執行一應用程式(APP)以顯示一操作介面11,該顯示介面係顯示伺服器10的資料或設定各模組的參數,如設定考試範圍及難易度,如圖3所示,使用者於該操作介面11設定考試範圍為六年級下學期的數學科,考試單元為第3與4章,測驗題數為5題,題型為自動組題,以及設定測驗時間為計時或不計時。 As shown in FIG. 2 and FIG. 3, the user executes an application (APP) through the electronic device 10A to display an operation interface 11, which displays the data of the server 10 or sets parameters of each module, such as setting the test range. And the difficulty level, as shown in FIG. 3, the user sets the examination scope to the mathematics of the sixth grade to the next semester in the operation interface 11, the examination unit is the third chapter and the fourth chapter, the test number is 5 questions, and the question type is the automatic group. Question, and set the test time to be timed or not.
本發明的運作流程可分為使用者端(電子裝置)與資料庫端(伺服器),使用者端係供使用者由電子裝置10A之操作介面11點觸操作,資料庫端係依所設參數由伺 服器10組題,依序進行下列步驟:使用者設定欲考試之試卷的範圍與鑑別度(101);判斷資料庫20之題庫內試題數量是否滿足考試之鑑別度所需試題數量(102),若題庫內試題數量足夠,則依所設之鑑別度直接組題並進行人工智慧驗證訓練(103),若題庫內試題數量不足,則隨機出題並標記結果將進行難易鑑別度訓練(104);使用者依據組題產生之考題進行考試,再將考試結果或答案送回資料庫(105);依照組題題目、答題成績與考生程度進行人工智慧訓練,使下次試卷的組題難度與鑑別度符合考生程度(106)。 The operation process of the present invention can be divided into a user end (electronic device) and a database end (server), and the user end is provided for the user to operate by the operation interface 11 of the electronic device 10A, and the database end is set according to the setting. Parameter by servo The server 10 groups of questions, the following steps are carried out in sequence: the user sets the scope and discriminating degree of the test paper to be tested (101); and determines whether the number of questions in the question bank of the database 20 satisfies the number of questions required for the examination of the test (102) If the number of questions in the question bank is sufficient, the questions are directly grouped according to the degree of discrimination and the artificial intelligence verification training is performed (103). If the number of questions in the question bank is insufficient, the questions are randomly selected and the results are marked for difficulty identification training (104) The user conducts the test according to the questions generated by the group questions, and then sends the test results or answers back to the database (105); according to the group title questions, the results of the questions and the degree of the candidates, the artificial intelligence training is performed, so that the difficulty of the group questions of the next test paper is The degree of discrimination is in accordance with the degree of candidate (106).
請參閱圖4、5所示,假設考生程度與考題難 度皆為常態分布,其可用來度量考生程度,如圖所示,學生人數為S1、S2至Sn,考生階級為Level1、Level2至Leveln,該等考生人數與考生階級作為支持向量機訓練的原始資料,以機率密度函數、標準差與變異數為參數,訓練該支持向量機並反覆驗證原始資料,調整參數與考生階級,使考生程度分佈為常態分布的態樣。例如以w * x-b=0,假設以支持向量機訓練後,若發現b=87.5,則w * x-87.5=0,即x=87.5(因為一維空間),因此,假設考試成績可用來區分考生程度高低,x=87.5可用來分類Level1的考生。同理,先以人工方式區分難易度,作為支持向量機訓練的原始資料,再如上所述將題目分類(假設考題難度亦為常態分布,造成訓練結果正確率太低,已可改用其他 分布方式),將考題難度與考生程度交叉驗證,確定高難度題目在高程度學生的階級下,分數為常態分布的訓練正確率有多高。或是將各種難度的題目平均出題,讓所有考生應試,檢查常態分布之訓練正確率,並反複驗證。 Please refer to Figures 4 and 5, assuming that the degree of candidates and the exam questions are difficult. Degrees are normal distribution, which can be used to measure the degree of candidates. As shown in the figure, the number of students is S1, S2 to Sn, and the candidate class is Level1, Level2 to Leveln. The number of candidates and the candidate class are the original training of support vector machine. The data, using probability density function, standard deviation and variance as parameters, train the support vector machine and repeatedly verify the original data, adjust the parameters and the candidate class, so that the degree of candidate distribution is the normal distribution. For example, w * xb = 0, assuming that after training with the support vector machine, if b = 87.5 is found, then w * x - 87.5 = 0, that is, x = 87.5 (because of the one-dimensional space), therefore, it is assumed that the test score can be used to distinguish The degree of candidates is high and low, x=87.5 can be used to classify Level1 candidates. In the same way, manually distinguish the difficulty level, as the original data of the support vector machine training, and then classify the questions as described above (assuming that the difficulty of the test questions is also normal distribution, resulting in a low rate of training results is too low, can be used to other Distribution method), cross-validation of the difficulty of the test questions and the degree of candidates, to determine how high the training accuracy rate of the high-skilled class is under the high-level student class. Or average the questions of all kinds of difficulty, let all candidates take the test, check the training accuracy rate of the normal distribution, and repeatedly verify.
由上述可知,利用人工智慧分析題庫中之各試 題的難度,依據所選鑑別度的高低,由大量題庫中挑選對應難度的試題並組成試卷,可達到最高鑑別度,使所有考生的分數呈現常態分布,而為一份有效區分各考生程度的試卷,測驗完畢後分析成績與考生程度並進行人工智慧訓練,使下次考試可準確地依照需求或所選鑑別度給與適當難度題目,並降低老師篩選考題時間,解決現有試卷由人工分類產生之鑑別度不佳的問題。 From the above, we can use artificial intelligence to analyze each test in the test bank. The difficulty of the question, according to the level of the selected degree of discrimination, the test questions of the corresponding difficulty are selected from a large number of test questions and composed of test papers, which can reach the highest degree of discrimination, so that the scores of all candidates are normally distributed, and the degree of each candidate is effectively distinguished. The test papers, after the test is completed, analyze the scores and the degree of candidates and carry out artificial intelligence training, so that the next test can accurately give appropriate difficulty questions according to the needs or the selected degree of discrimination, and reduce the teacher's screening test time, and solve the existing test papers generated by manual classification. The problem of poor discrimination.
10‧‧‧伺服器 10‧‧‧Server
20‧‧‧資料庫 20‧‧‧Database
30‧‧‧題目難易鑑別度模組 30‧‧‧Difficult identification module
40‧‧‧題目篩選模組 40‧‧‧Title screening module
50‧‧‧考試結果驗證模組 50‧‧‧Exam results verification module
60‧‧‧考生程度判別模組 60‧‧‧ Candidate Level Discrimination Module
Claims (6)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW102116514A TW201443667A (en) | 2013-05-09 | 2013-05-09 | Artificial intelligent test paper item system and item method thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW102116514A TW201443667A (en) | 2013-05-09 | 2013-05-09 | Artificial intelligent test paper item system and item method thereof |
Publications (1)
Publication Number | Publication Date |
---|---|
TW201443667A true TW201443667A (en) | 2014-11-16 |
Family
ID=52423347
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW102116514A TW201443667A (en) | 2013-05-09 | 2013-05-09 | Artificial intelligent test paper item system and item method thereof |
Country Status (1)
Country | Link |
---|---|
TW (1) | TW201443667A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104504953A (en) * | 2014-12-30 | 2015-04-08 | 浪潮(北京)电子信息产业有限公司 | Method and device for randomly generating examination paper |
CN110083816A (en) * | 2019-05-13 | 2019-08-02 | 北京北师智慧科技有限公司 | Digitalized teaching method, device and its system |
TWI731397B (en) * | 2018-08-24 | 2021-06-21 | 宏達國際電子股份有限公司 | Method for verifying training data, training system, and computer program product |
-
2013
- 2013-05-09 TW TW102116514A patent/TW201443667A/en unknown
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104504953A (en) * | 2014-12-30 | 2015-04-08 | 浪潮(北京)电子信息产业有限公司 | Method and device for randomly generating examination paper |
TWI731397B (en) * | 2018-08-24 | 2021-06-21 | 宏達國際電子股份有限公司 | Method for verifying training data, training system, and computer program product |
CN110083816A (en) * | 2019-05-13 | 2019-08-02 | 北京北师智慧科技有限公司 | Digitalized teaching method, device and its system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Juhaňák et al. | Using process mining to analyze students' quiz-taking behavior patterns in a learning management system | |
Mao et al. | Validation of automated scoring for a formative assessment that employs scientific argumentation | |
KR100978091B1 (en) | Management system for online test assessment and method thereof | |
US20080057480A1 (en) | Multimedia system and method for teaching basal math and science | |
Nuhfer et al. | Random number simulations reveal how random noise affects the measurements and graphical portrayals of self-assessed competency | |
Nichols et al. | Standard‐setting methods as measurement processes | |
CN107657559A (en) | A kind of Chinese reading capability comparison method and system | |
CN103577507A (en) | Intelligent question bank system with real-time detection and self-adaptive evolution mechanism and method | |
Liu et al. | Collaborative learning team formation: a cognitive modeling perspective | |
CN110909035A (en) | Personalized review question set generation method and device, electronic equipment and storage medium | |
Dowd et al. | Concept maps for improved science reasoning and writing: Complexity isn’t everything | |
CN108009761A (en) | Teaching quality management method and system | |
Hardegree | Standards-based assessment and high stakes testing: Accuracy of standards-based grading | |
Clarke et al. | Classification accuracy of easyCBM first-grade mathematics measures: Findings and implications for the field | |
McGrane et al. | Applying a thurstonian, two-stage method in the standardized assessment of writing | |
KR20140127483A (en) | System of a sham examination and study using application | |
CN114330997A (en) | Intelligent teaching plan generating system based on BOPPPS model | |
TW201443667A (en) | Artificial intelligent test paper item system and item method thereof | |
CN114140280A (en) | Data processing method and system based on AI correction and electronic equipment | |
Wu et al. | Establishing the criterion-related, construct, and content validities of a simulation-based assessment of inquiry abilities | |
TWI496121B (en) | Evaluation System and Its Method of Networked Implementation | |
Ortega-Sanchez | Written exams: How effectively are we using them? | |
Hambleton et al. | Solving criterion-referenced measurement problems with item response models | |
Duffy et al. | Measurement of first year engineering students cognitive activities using a spatial skills test and an electrical concepts test: implications for curriculum design | |
Tapsai | The Online Test Bank Management System in Integration Model |