An index establishing method and device for test questions and electronic equipment

Abstract

The present application relates to the technical field of intelligent retrieval, in particular to an index establishing method and device for test question retrieval and electronic equipment, the index establishing method for test question retrieval comprises: obtaining test question resources, judging test question attributes; according to the test question attributes of the test question resources, a preset text preprocessing strategy corresponding to the test question attributes is used to establish an index. The index establishing method for test question retrieval according to the test question attributes of the test question retrieval object, that is, the test questions are divided into subjects, such as Chinese test questions, mathematics test questions, English test questions, etc.; the test questions are also divided into types, such as reading questions, which are specifically established indexes, realizing an index establishing method for test question retrieval by subject and / or type, and using a retrieval strategy corresponding to the index establishment when retrieving test questions, which can effectively improve the retrieval efficiency of test question retrieval and the matching degree of retrieval results.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent retrieval, in particular to an index establishing method and device for test question retrieval and electronic equipment. BACKGROUND

[0002] Retrieval, simply speaking, is a process of searching for required information from information collections such as literature and network information.

[0003] Intelligent retrieval is developed from word extraction retrieval and full-text retrieval, which is a multimedia retrieval system with high judgment, understanding and processing capabilities for retrieval words. The existing intelligent retrieval technology mainly includes text intelligent retrieval technology, image intelligent retrieval technology and video intelligent retrieval technology. With the development of the times, there are more and more information based on web, and how to obtain the information really needed in the vast amount of information has become a great challenge.

[0004] When obtaining information, the corresponding time of sequential search will become unbearable. The solution to the search response time is to preprocess the files in the text document library and establish a data structure index for the text library, which is convenient for searching. The search technology based on index is very suitable for large-scale, stable or medium-term changes in text document library, and most search engines use index-based retrieval technology. Therefore, the establishment of index is directly related to the efficiency of retrieval and the relevance of search results, and is the basis of intelligent retrieval technology.

[0005] In the field of online education, users can obtain relevant information of search test questions (such as test question analysis, knowledge points involved in test questions, test papers and years to which test questions belong, etc.) through test question retrieval technology. Since test question retrieval has a clear purpose of test questions, how to establish index according to the characteristics of test questions is crucial to the speed of test question retrieval and the matching degree of retrieval results.

[0006] Therefore, the present application aims to solve the problem of index establishment for test question retrieval. SUMMARY

[0007] In order to solve the above problems, the present application provides an index establishing method for test question retrieval, which comprises:

[0008] Obtaining test question resources and judging test question attributes;

[0009] Establishing index according to the test question attributes of test question resources by using a preset text preprocessing strategy corresponding to the test question attributes.

[0010] As an optional embodiment of the present application, the test question resource attributes include subject information.

[0011] According to the subject information of the test question resource, a preset text preprocessing strategy corresponding to the subject is adopted to establish the index, including:

[0012] The test question resource is a science test question, after the test question resource is segmented, the formula in the test question resource is extracted, and the formula is used as an index word to establish an index alone;

[0013] Optionally, the subject symbols related to the subject of the test question are reserved before the test question resource is segmented, and the subject symbols are normalized. As an optional embodiment of the application, the formula in the test question resource is extracted, and the formula is used as an index word to establish an index alone, including:

[0014] The extracted formula in the test question resource is converted into a unified set format, and the formula in the unified format is used as an index word to establish an index alone;

[0015] Optionally, the extracted formula in the test question resource is in LaTeX format, and is converted into an OCR recognizable formula string through format conversion.

[0016] As an optional embodiment of the application, the formula in the test question resource is extracted, and the formula is used as an index word to establish an index alone, including:

[0017] An easy-to-recognize error automatic correction operation is performed on the extracted formula in the test question resource, the corrected formula is used as an index word to establish an index alone, and the original formula in the extracted test question resource is used as an index word to establish a redundant index.

[0018] As an optional embodiment of the application, the test question resource attribute includes subject information;

[0019] According to the subject information of the test question resource, a preset text preprocessing strategy corresponding to the subject is adopted to establish the index, including:

[0020] The test question is a test question suitable for low and young age groups, after the test question resource is segmented, the test question picture in the test question resource is extracted, and the test question picture is used to establish an index.

[0021] As an optional embodiment of the application, the test question resource attribute includes question type information;

[0022] According to the question type information of the test question resource, a preset text preprocessing strategy corresponding to the question type is adopted to establish the index, including:

[0023] The test question is a reading test question, the stem part of the test question resource is extracted and compressed and fragmented, and the compressed stem fragment is used to establish an index.

[0024] As an optional embodiment of the application, the test question resource is obtained, and the test question attribute is judged, including:

[0025] Obtaining a test question resource, extracting a string in the test question resource, judging a subject to which the test question belongs according to a proportion of a Chinese string, and / or an English string, and / or a numerical string;

[0026] And / or obtaining the test question resource, judging whether the test question resource contains a keyword and / or a key word corresponding to a type of the test question, so as to judge the type of the test question.

[0027] As an optional embodiment of the present application, a duplicate judgment strategy is executed before the index is established based on the obtained test question resource, and the duplicate judgment strategy comprises:

[0028] Judging a subject to which the test question resource belongs, generating a retrieval word segmentation according to a text preprocessing strategy corresponding to the subject, performing a retrieval data recall in an index library according to the retrieval word segmentation, comparing the retrieval data with test question resource data, and feeding back a duplicate test question, a non-duplicate test question and a duplicate judgment comparison result to be manually confirmed.

[0029] The present application also provides an index establishment device for test question retrieval, comprising:

[0030] A test question judgment module obtains a test question resource and judges a test question attribute;

[0031] An index establishment module establishes an index according to the test question attribute of the test question resource by using a preset text preprocessing strategy corresponding to the test question attribute.

[0032] The present application also provides an electronic device comprising a processor and a memory, wherein the memory is used to store a computer executable program, and when the computer executable program is executed by the processor, the processor executes an index establishment method for test question retrieval according to any one of claims 1-8.

[0033] Compared with the prior art, the present application has the following beneficial effects:

[0034] The index establishment method for test question retrieval according to the present application establishes an index according to a test question attribute of a test question retrieval object, that is, the test question is divided into subjects such as Chinese test questions, mathematics test questions and English test questions, and the test question is also divided into types such as reading questions, so that the index establishment method for test question retrieval according to the present application is divided into subjects and / or types, a retrieval strategy corresponding to the index establishment is used when the test question is retrieved, and the retrieval efficiency of the test question retrieval and the matching degree of the retrieval result can be effectively improved.

[0035] The index establishment method for test question retrieval according to the present application is aimed at the characteristics of science test questions, retains a formula in the test question, and establishes an index by using the formula as an index word, so that the key part, that is, the formula in the science test question is retained and indexed.

[0036] The application provides a way of establishing index according to the type of test questions, since reading test questions, whether Chinese reading test questions or English reading test questions, are based on the whole article, therefore, for reading test questions, the whole article does not need to be segmented and indexed, but the whole stem is segmented (the whole article) to establish index, which greatly improves the efficiency of establishing index of reading test questions, and the matching degree of the search results is higher when searching reading test questions. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 Flow chart of the index establishing method of the embodiment of the application;

[0038] Figure 2 Flow chart of the index establishing method of the embodiment of the application for science test questions;

[0039] Figure 3 Flow chart of the index establishing method of the embodiment of the application for English test questions;

[0040] Figure 4 Flow chart of the index establishing method of the embodiment of the application for other arts test questions;

[0041] Figure 5 Flow chart of the index establishing method of the embodiment of the application for test questions suitable for low and young age groups;

[0042] Figure 6 Flow chart of the index establishing method of the embodiment of the application for reading test questions. DETAILED DESCRIPTION

[0043] In order to make the objects, technical solutions and advantages of the embodiments of the application clearer, the technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments of the application.

[0044] Therefore, the detailed description of the embodiments of the application below is not intended to limit the scope of the claimed application, but only represents some embodiments of the application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the application without creative labor are within the scope of protection of the application.

[0045] It should be noted that the embodiments in the application and the features and technical solutions in the embodiments can be combined with each other without conflict.

[0046] It should be noted that: similar labels and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0047] In the description of the present application, it should be noted that the terms "upper", "lower", and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship commonly used when the product of the present application is used, or the orientation or positional relationship commonly understood by those skilled in the art. Such terms are only for the convenience of describing the present application and simplifying the description, and are not intended to indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second", and the like are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0048] Referring to Figure 1 The embodiment provides an index establishing method for test question retrieval, which comprises the following steps:

[0049] Obtaining test question resources and judging test question attributes;

[0050] According to the test question attributes of the test question resources, a preset text preprocessing strategy corresponding to the test question attributes is used to establish an index.

[0051] The index establishing method for test question retrieval according to the test question attributes of the test question retrieval object, i.e., test questions are divided into subjects, such as Chinese test questions, mathematics test questions, and English test questions, and test questions are also divided into types, such as reading questions, which are used to establish an index, so that a subject-divided and / or type-divided index establishing method for test question retrieval is realized. When test questions are retrieved, a retrieval strategy corresponding to the index establishment is used, so that the retrieval efficiency of test question retrieval and the matching degree of the retrieval result can be effectively improved.

[0052] The index establishing method for test question retrieval according to the preset subject-divided and / or type-divided text preprocessing strategy. When test questions are established, the subject and / or type to which the test question resources belong are first judged, and then the subject-divided and / or type-divided text preprocessing strategy is used to establish an index.

[0053] Specifically, referring to Figure 2 As an optional embodiment of the present embodiment, the test question resource attributes described in the present embodiment include subject information.

[0054] According to the subject information of the test question resources, a preset text preprocessing strategy corresponding to the subject is used to establish an index, which comprises the following steps:

[0055] The test question resources are scientific test questions. After the test question resources are segmented, formulas in the test question resources are extracted, and the formulas are used as index words to establish an index.

[0056] The science test questions of the embodiment include, but are not limited to, mathematical test questions, physical test questions and chemical test questions. Since the science test questions generally contain formulas, and the formulas are generally the key to solving the problems, the existing index establishing method focuses on the key words and the word reservation. The index establishing method of the embodiment reserves the formulas in the test questions according to the characteristics of the science test questions, and establishes the index with the formulas as the index words, so as to ensure that the key part, i.e. the formulas, in the science test questions are reserved and indexed.

[0057] Optionally, the subject symbols related to the subject of the test questions are reserved before the test questions are segmented, and the subject symbols are normalized. Since the input difference of the formulas causes the inconsistency of the subject symbols, such as the mathematical subject symbols of "x" and "*", " / ", and fraction representation, the normalization rules are preset to unify the subject symbols, and then the index is established, so as to avoid the complexity of the index establishment and ensure the accuracy of the index establishment.

[0058] Further, the embodiment extracts the formulas in the test question resources, and establishes the index with the formulas as the index words.

[0059] The extracted formulas in the test question resources are converted into a unified set format, and then the formulas in the unified format are used as the index words to establish the index.

[0060] Since the formulas are generally edited by a formula editor, the formulas generated by the formula editor may have different formats. Therefore, the science formulas in different formats are converted and unified before the index is established with the formulas as the index words.

[0061] Optionally, the extracted formulas in the test question resources are in LaTeX format, and are converted into an OCR recognizable formula string through the format conversion. This is mainly because the test question retrieval is mainly test question photograph search, and the test question content is extracted through the OCR recognition technology. Therefore, the formulas are converted into the OCR recognizable formula string to establish the index, which is more suitable for the test question retrieval of the user.

[0062] Further, the embodiment extracts the formulas in the test question resources, and establishes the index with the formulas as the index words.

[0063] The extracted formulas in the test question resources are subjected to an easy-to-recognize error automatic correction operation, the corrected formulas are used as the index words to establish the index, and the original formulas in the extracted test question resources are used as the index words to establish a redundant index.

[0064] Since the formula contains letters, symbols, etc., there may be obvious recognition errors when extracting from the test question resource, such as recognizing "+" as "ten", recognizing "-" as "one", and recognizing "O" as "mouth". When such obvious errors are recognized, correction operation is needed to ensure the accuracy of the formula and the accuracy of the index establishment. In addition, in order to prevent the misoperation caused by automatic correction, the original formula in the extracted test question resource is used to establish a redundant index as an index word.

[0065] Specifically, the index establishment method of the test question of the present embodiment includes:

[0066] When the test question is a mathematics test question, the mathematical symbols in the test question are retained, the mathematical formula is extracted, the mathematical formula is automatically corrected for easy recognition errors, and the mathematical formula is indexed separately.

[0067] When the test question is a physics test question, the physical symbols in the test question are retained, the physical formula is extracted, the physical formula is automatically corrected for easy recognition errors, and the physical formula is indexed separately.

[0068] When the test question is a chemistry test question, the chemical symbols in the test question are retained, the chemical formula is extracted, the chemical formula is automatically corrected for easy recognition errors, and the chemical formula is indexed separately.

[0069] As an optional implementation of the present embodiment, the present embodiment establishes an index according to the subject of the test question resource using a preset text preprocessing strategy of the corresponding subject, which includes:

[0070] Referring to Figure 5 As shown in the figure, the present embodiment extracts the test question picture in the test question resource after word segmentation of the test question, and indexes the test question picture.

[0071] Since the test question suitable for low and young age groups is mostly a picture question, the picture index can be established according to the test question picture for the test question suitable for low and young age groups, supporting picture retrieval.

[0072] Referring to Figure 4 As an optional implementation of the present embodiment, the present embodiment establishes an index according to the subject of the test question resource using a preset text preprocessing strategy of the corresponding subject, which includes:

[0073] The test question is other liberal arts test questions, the special characters in the test question resource are deleted, the word segmentation is performed according to Chinese semantics, and the index is established according to the items of word segmentation. The other liberal arts of the present embodiment include politics, history, etc.

[0074] Referring to Figure 3As an optional implementation of the present embodiment, the present embodiment discloses a method for establishing an index according to a subject of a test question resource, which comprises the following steps:

[0075] The test question is an English test question, spaces are reserved, and the test question is segmented according to spaces, and an index is established according to segmented items.

[0076] Referring to Figure 6 As an optional implementation of the present embodiment, the present embodiment discloses a test question resource attribute, which comprises question type information.

[0077] According to the question type information of the test question resource, a preset text preprocessing strategy corresponding to the question type is adopted to establish an index, which comprises the following steps: the test question is a reading test question, a stem part of the test question resource is extracted and compressed and segmented, and an index is established according to the compressed and segmented stem.

[0078] The present embodiment provides a method for establishing an index according to a question type of a test question. Since a reading test question is based on a whole article, whether the reading test question is a Chinese reading test question or an English reading test question, the whole article is segmented, and then an index is established. For the reading test question, the whole article is segmented, and then an index is established. The efficiency of establishing an index of the reading test question is greatly improved, and the matching degree of a search result is higher when the reading test question is searched.

[0079] As an optional implementation of the present embodiment, the present embodiment discloses a method for obtaining a test question resource and judging a subject thereof, which comprises the following steps:

[0080] The test question resource is obtained, a string in the test question resource is extracted, and a subject of the test question is judged according to a proportion of a Chinese string, and / or an English string, and / or a numerical string;

[0081] and / or the test question resource is obtained, and it is judged whether the test question resource contains a keyword or a key word corresponding to a question type, so as to judge a question type of the test question.

[0082] Specifically, the proportion of the string can be set. When the proportion of the subject string exceeds the set proportion, the subject can be determined. Whether the test question resource contains a keyword of a reading question can be judged to determine whether the test question is a reading question type.

[0083] As an optional implementation of the present embodiment, the present embodiment discloses a method for establishing an index for a test question search. If the obtained test question resource cannot judge a subject and / or a question type thereof, an index is created in a conventional segmentation manner.

[0084] The method for establishing an index for a test question search in the present embodiment comprises the following steps of creating an index in a conventional segmentation manner:

[0085] 1. Obtain test questions from a test question resource library, which can be submitted by users or actively search for test questions with low matching degrees.

[0086] 2. Extract text based on the test question resource library, mainly to extract strings of various formats.

[0087] 3. After extracting the text string, the text string needs to be preprocessed to select appropriate objects to establish an index.

[0088] Text preprocessing first analyzes the words contained in the text, i.e. word segmentation. In terms of semantic expression, not all words have the same expression ability, so in addition to word segmentation, text preprocessing also includes stop word deletion, stem extraction, index word selection, and dictionary building.

[0089] 3.1 Word segmentation

[0090] Word segmentation technology belongs to the field of natural language processing. For a sentence, a person can understand which are words and which are not by their own instructions, but how can a computer also understand it? The processing process is the word segmentation algorithm. Common word segmentation methods include: (1) string matching-based word segmentation method - forward maximum matching algorithm based on string matching; (2) statistical-based word segmentation method - also known as dictionary-free word segmentation or statistical word selection method; (3) understanding-based word segmentation method - experimental stage.

[0091] The forward maximum matching algorithm based on string matching, also known as the "mechanical word segmentation method", matches the Chinese character string to be analyzed with a "sufficiently large" machine dictionary according to a certain strategy (algorithm). If a string is found in the dictionary, the matching is successful (a word is recognized).

[0092] According to the different scanning directions, the mechanical word segmentation method can be divided into forward matching and reverse matching; according to the different length priority matching conditions, it can be divided into maximum (longest) matching and minimum (shortest) matching.

[0093] For the mechanical word segmentation method, a general model, ASM, can be established. The model form is identified as ASM(d, a, m), where d, a and m have the following meanings and values:

[0094] d: matching direction, +1 for forward direction, -1 for reverse direction.

[0095] a: increase or decrease the length of the string (number of characters) after each matching failure, +1 for increasing, -1 for decreasing;

[0096] m: maximum or minimum matching flag, +1 for maximum matching, -1 for minimum matching.

[0097] In practical application, the forward maximum matching algorithm based on string matching [ASM(+1, -1, +1)] is a widely used mechanical word segmentation method, which is based on only one word segmentation table and a basic segmentation evaluation principle (i.e. the "long word first" principle) to perform word segmentation. This segmentation method requires the least language materials, simple program and short development cycle.

[0098] 3.2 Stop word deletion

[0099] Generally, the words appearing in the text resource library with a frequency of more than 80% do not work for the retrieval process at all, and this part of the words is called stop words.

[0100] When selecting the index words, stop words need to be filtered to improve the index efficiency. Generally, articles, prepositions and conjunctions are stop words, and in fact, the retrieval system will set a stop word list. Deleting stop words can greatly reduce the size of the index space.

[0101] 3.3 Stemming

[0102] Stemming refers to the part left after the affix of a word is deleted, and stemming is an operation taken to solve the problems existing in English retrieval. In English retrieval, if the word input by the user is a variant of a word in a related text in the information base, other forms are considered irrelevant texts, which greatly affects the recall rate.

[0103] In order to solve this problem, the stem is used to replace all variants of the stem when building the index. In this way, not only the recall rate can be improved and the performance of information retrieval can be improved, but also the vocabulary size of the constructed index will be greatly reduced, and the index space will be further reduced.

[0104] 3.4 Selection of index words

[0105] If full-text indexing is used, all words in the library are indexed, and indexing some words with weak semantic expression ability will waste the system's index space and affect the system's retrieval performance, so not all words appearing in the document are indexed, but some important words are selected to be indexed.

[0106] 3.5 Building a dictionary

[0107] Dictionary refers to a thesaurus or a classified compilation, and in a text retrieval system, a dictionary is a data compilation used to find related word information according to a word.

[0108] 4. Indexing

[0109] The result of the text preprocessing will be used to build the core of the text retrieval system, the index. The data structure of the index is inverted file, inverted file or inverted index, inverted list, which is used to improve the query speed. The inverted file consists of two parts, the vocabulary and the word occurrence. The vocabulary is generally stored in a special data structure to improve the query speed of the word. For each word in the vocabulary, there is a list in the word occurrence to record the position of the word in the text. As an optional implementation of the embodiment, the index establishment method for test question retrieval of the embodiment executes the duplicate judgment strategy before starting to establish the index after obtaining the test question resource. The duplicate judgment strategy includes:

[0110] judging the subject of the test question resource, and generating retrieval words according to the text preprocessing strategy corresponding to the subject, retrieving data from the index library according to the retrieval words, comparing the retrieved data with the test question resource data, and feeding back the duplicate test question, non-duplicate test question and manual confirmation judgment comparison results.

[0111] When the duplicate judgment comparison result of the retrieved data and the test question resource data is that the coincidence rate is above the first set value, the duplicate test question duplicate judgment result is fed back.

[0112] When the duplicate judgment comparison result of the retrieved data and the test question resource data is that the coincidence rate is below the second set value, the non-duplicate test question duplicate judgment result is fed back.

[0113] When the duplicate judgment comparison result of the retrieved data and the test question resource data is that the coincidence rate is between the second set value and the first set value, the manual confirmation judgment comparison result is fed back, and the retrieved data is sent to the relevant personnel for manual duplicate judgment comparison.

[0114] Optionally, the first set value is 90%, and the second set value is 80%.

[0115] The embodiment also provides an index establishment device for test question retrieval, which comprises:

[0116] a test question judgment module, which obtains test question resources and judges test question attributes;

[0117] an index establishment module, which establishes an index according to the test question attributes of the test question resources by using a preset text preprocessing strategy corresponding to the test question attributes.

[0118] The index establishing device for test question retrieval of the embodiment is established according to the characteristics of the test question retrieval object, that is, test questions are divided into subjects, such as Chinese test questions, mathematics test questions, English test questions, etc.; test questions are also divided into types, such as reading questions, and an index establishing method for test question retrieval is realized according to subjects and / or types. When test question retrieval is performed, a retrieval strategy corresponding to the index establishing is adopted, and the retrieval efficiency of test question retrieval and the matching degree of the retrieval result can be effectively improved.

[0119] The index establishing device for test question retrieval of the embodiment is preset with a text preprocessing strategy of a corresponding subject and / or type. When the index of the test question is established, the subject and / or type to which the test question resource belongs is first judged, and then the index is established according to the preset text preprocessing strategy of the corresponding subject and / or type.

[0120] Specifically, as an optional implementation of the embodiment, the index establishing module establishes the index according to the subject of the test question resource by using the preset text preprocessing strategy of the corresponding subject, which includes:

[0121] The test question judging module judges that the test question resource is a science test question. After the index establishing module performs word segmentation on the test question resource, the formula in the test question resource is extracted, and the index is established by taking the formula as an index word.

[0122] The science test question of the embodiment includes but is not limited to mathematics test questions, physics test questions, and chemistry test questions. Since the science test question generally contains a formula, and the formula is generally the key to solving the problem, the existing index establishing method mainly focuses on the retention of keywords and words. The index establishing device of the embodiment retains the formula in the test question according to the characteristics of the science test question, and establishes the index by taking the formula as an index word, so that the key part, that is, the formula, in the science test question is retained and indexed.

[0123] Optionally, after the test question resource is segmented, the subject symbols related to the subject of the test question are retained, and the subject symbols are normalized. Due to the difference in the input of the formula, the subject symbols are inconsistent, such as “×” and “*”, “÷” and “ / ”, and fractional representation in the mathematics subject. The normalization rule is preset, and the index is established again to avoid the complexity of index establishment and ensure the accuracy of index establishment.

[0124] Further, the index establishing module extracts the formula in the test question resource and establishes the index by taking the formula as an index word, which includes:

[0125] The index establishing module converts the format of the extracted formula in the test question resource into a unified set format, and then establishes the index by taking the formula in the unified format as an index word.

[0126] Since the formula is generally edited by the formula editor, the formula edited by the formula editor can have different formats, and therefore, the scientific formulas of different formats need to be converted and unified before indexing with the formula as the index word.

[0127] Optionally, the extracted formula in the test question resource is in LaTeX format, and is converted into an OCR recognizable formula string through format conversion. This is mainly because the test question retrieval mode is mainly test question photograph search, and the test question photo is extracted through OCR recognition technology. Therefore, the formula is indexed as an OCR recognizable formula string, which is more suitable for user test question retrieval.

[0128] Further, the index establishing module extracts the formula in the test question resource, and separately establishes an index with the formula as the index word, including:

[0129] The extracted formula in the test question resource is subjected to automatic correction of easily recognizable errors, and an index is separately established with the corrected formula as the index word, and a redundant index is established with the original formula in the extracted test question resource as the index word.

[0130] Since the formula contains letters, symbols, etc., there can be obvious recognition errors when extracting from the test question resource, such as recognizing "+" as "ten", "-" as "one", and "O" as "mouth". When such obvious errors are recognized, correction operation is needed to ensure the accuracy of the formula and the accuracy of the index establishment. In addition, in order to prevent the automatic correction from causing misoperation, the embodiment also establishes a redundant index with the original formula in the extracted test question resource as the index word.

[0131] Specifically, the index establishing module for the science test question includes:

[0132] When the test question judging module judges that the test question is a mathematical test question, the index establishing module retains the mathematical symbols in the test question, extracts the mathematical formula, automatically corrects the easily recognizable errors of the mathematical formula, and separately establishes an index for the mathematical formula.

[0133] When the test question judging module judges that the test question is a physical test question, the index establishing module retains the physical symbols in the test question, extracts the physical formula, automatically corrects the easily recognizable errors of the physical formula, and separately establishes an index for the physical formula.

[0134] When the test question judging module judges that the test question is a chemical test question, the index establishing module retains the chemical symbols in the test question, extracts the chemical formula, automatically corrects the easily recognizable errors of the chemical formula, and separately establishes an index for the chemical formula.

[0135] As an optional implementation of the embodiment, the index establishing module establishes the index according to the preset text preprocessing strategy of the corresponding subject according to the subject of the test question resource, and the index establishing module includes:

[0136] The test question judgment module judges that the test question is an infant test question, the index establishing module extracts the test question picture in the test question resource after the test question resource is segmented, and the index is established according to the test question picture.

[0137] Since the infant test question is mostly a picture question, the picture index can be established according to the test question picture for the infant test question, and the picture retrieval is supported.

[0138] As an optional implementation of the embodiment, the index establishing module establishes the index according to the preset text preprocessing strategy of the corresponding subject according to the subject of the test question resource, and the index establishing module includes:

[0139] The test question judgment module judges that the test question is an other liberal arts test question, the index establishing module deletes the special characters in the test question resource, segments the Chinese semantics, and establishes the index according to the segmented items.

[0140] As an optional implementation of the embodiment, the index establishing module establishes the index according to the preset text preprocessing strategy of the corresponding subject according to the subject of the test question resource, and the index establishing module includes:

[0141] The test question judgment module judges that the test question is an English test question, the index establishing module retains the space, segments the space, and establishes the index according to the segmented items.

[0142] As an optional implementation of the embodiment, the index establishing module establishes the index according to the preset text preprocessing strategy of the corresponding subject according to the subject of the test question resource, and the index establishing module includes:

[0143] The test question judgment module judges that the test question is a reading test question, the index establishing module extracts the stem part of the test question resource after the test question resource is segmented, and establishes the index according to the compressed fragments of the stem.

[0144] The embodiment provides a way of establishing an index according to a test question type. Since the reading test question is based on a whole article, whether it is a Chinese reading test question or an English reading test question, the whole article is segmented for the reading test question, and then the index is established, but the whole stem fragment (the whole article) is established as the index, which greatly improves the efficiency of establishing the index of the reading test question, and the matching degree of the retrieval result is higher when the reading test question is retrieved.

[0145] As an optional implementation of the embodiment, the test question judgment module obtains the test question resource, and judges the subject to which the test question resource belongs, and the test question judgment module includes:

[0146] The test question judgment module obtains a test question resource, extracts a character string in the test question resource, and judges a subject to which the test question belongs according to a proportion of a Chinese character string, and / or an English character string, and / or a numerical character string;

[0147] The test question judgment module obtains a test question resource, judges whether the test question resource contains a keyword or a key word corresponding to a question type, so as to judge the question type to which the test question belongs.

[0148] Specifically, the proportion of the character string can be set, and when the proportion of the subject character string exceeds the set proportion, the subject to which the test question belongs can be determined. Whether the test question is a reading question type can be judged by judging whether the test question resource contains a keyword of a reading question.

[0149] As an optional implementation of the embodiment, an index establishment device for test question retrieval of the embodiment includes a duplicate judgment module that executes a duplicate judgment strategy before starting to establish an index on the basis of obtaining a test question resource. The duplicate judgment strategy includes:

[0150] judging a subject to which the test question resource belongs, generating retrieval word segmentation according to a text preprocessing strategy of the corresponding subject, performing retrieval data recall in an index library according to the retrieval word segmentation, comparing the retrieval data and test question resource data, and feeding back a duplicate test question, a non-duplicate test question, and a duplicate comparison result to be manually confirmed.

[0151] When the duplicate comparison result of the retrieval data and the test question resource data is that a coincidence rate is above a first set value, a duplicate test question duplicate comparison result is fed back.

[0152] When the duplicate comparison result of the retrieval data and the test question resource data is that the coincidence rate is below a second set value, a non-duplicate test question duplicate comparison result is fed back.

[0153] When the duplicate comparison result of the retrieval data and the test question resource data is that the coincidence rate is between the second set value and the first set value, a duplicate comparison result to be manually confirmed is fed back, and a retrieval data recall result is sent to relevant personnel for manual duplicate comparison.

[0154] Optionally, the first set value is 90%, and the second set value is 80%.

[0155] The embodiment further provides an electronic device including a processor and a memory. The memory is used to store a computer executable program. When the computer executable program is executed by the processor, the processor executes the index establishment method for test question retrieval.

[0156] The embodiment further provides a computer readable storage medium that stores a computer executable program. When the computer executable program is executed, the index establishment method for test question retrieval is implemented.

[0157] From the above description of the embodiments, it is easy for those skilled in the art to understand that the present application can be implemented by hardware capable of executing a specific computer program, such as a system of the present application and an electronic processing unit, a server, a client, a mobile phone, a control unit, a processor, etc. contained in the system. The present application can also be implemented by computer software for executing the method of the present application, such as control software executed by a microprocessor, an electronic control unit, a client, a server, etc. It should be noted that the computer software for executing the method of the present application is not limited to being executed by one or specific hardware entity, but can also be implemented in a distributed manner by non-specific hardware. For computer software, the software product can be stored in a computer-readable storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.), or can be distributed on a network, as long as it can enable an electronic device to execute the method according to the present application.

[0158] The above embodiments are only used to illustrate the present application and not to limit the technical solutions described in the present application. Although the present application has been described in detail with reference to the above embodiments, the present application is not limited to the above specific embodiments, and any modification or equivalent replacement of the present application; all technical solutions and improvements that do not deviate from the spirit and scope of the present application are encompassed in the scope of the claims of the present application.

Claims

1. An index creation method for search of test questions, characterized by, The method comprises the following steps: Obtaining a test question resource and judging a test question attribute; Indexing according to a preset text preprocessing strategy corresponding to the test question attribute of the test question resource; The test question attribute of the test question resource comprises subject information, and indexing according to a preset text preprocessing strategy corresponding to the subject according to the subject information of the test question resource comprises: The test question resource is a science test question, the test question resource is segmented, a formula in the test question resource is extracted, and indexing is separately performed on the formula as an index word; The extracting of the formula in the test question resource and the indexing on the formula as an index word comprise: converting the extracted formula in the test question resource into a uniform set format, and then indexing on the formula in the uniform format as an index word; The extracted formula in the test question resource is in LaTeX format, and is converted into an OCR recognizable formula string through format conversion; The extracting of the formula in the test question resource and the indexing on the formula as an index word comprise: performing automatic correction of easily recognizable errors on the extracted formula in the test question resource, indexing on the corrected formula as an index word, and simultaneously indexing on the original formula in the test question resource as an index word; When the test question is a mathematics test question, a mathematical symbol in the test question is retained, a mathematical formula is extracted, the mathematical formula is automatically corrected for easily recognizable errors, and the mathematical formula is indexed separately; When the test question is a physics test question, a physics symbol in the test question is retained, a physics formula is extracted, the physics formula is automatically corrected for easily recognizable errors, and the physics formula is indexed separately; When the test question is a chemistry test question, a chemistry symbol in the test question is retained, a chemistry formula is extracted, the chemistry formula is automatically corrected for easily recognizable errors, and the chemistry formula is indexed separately; The test question attribute comprises subject information, and indexing according to a preset text preprocessing strategy corresponding to the subject according to the subject information of the test question resource comprises: The test question is a test question suitable for a low and young age group, the test question resource is segmented, a test question picture in the test question resource is extracted, and indexing is performed on the test question picture; The indexing according to the preset text preprocessing strategy corresponding to the subject according to the subject information of the test question resource comprises: The test question is another liberal arts test question, special characters in the test question resource are deleted, segmentation is performed according to Chinese semantics, and indexing is performed according to items of the segmentation; The indexing according to the preset text preprocessing strategy corresponding to the subject according to the subject information of the test question resource comprises: The test question is an English test question, spaces are retained, segmentation is performed according to the spaces, and indexing is performed according to items of the segmentation; The test question attribute comprises question type information, and indexing according to a preset text preprocessing strategy corresponding to the question type according to the question type information of the test question resource comprises: The test question is a reading test question, a stem part of the test question resource is extracted and compressed into fragments, and indexing is performed on the compressed stem fragments; Before the segmentation of the test question resource, subject symbols related to a subject of the test question are retained, and the subject symbols are normalized.

2. The index creation method for test question search according to claim 1, wherein, The obtaining of the test question resource and the judging of the test question attribute comprise: Obtaining a test question resource, extracting character strings within the test question resource, and determining the subject to which the test question belongs based on the ratio of Chinese character strings, and / or English character strings, and / or numeric character strings; And / or obtain test question resources to determine whether they contain keywords or key words of the corresponding question type, so as to determine the question type to which the test question belongs.

3. The method of claim 1 or 2, wherein the index is built for each of the test questions. Before obtaining the test question resources and starting to build the index, the duplicate detection strategy is executed. The duplicate detection strategy includes: Determine the subject to which the test question resources belong, and generate search terms according to the text preprocessing strategy of the corresponding subject. Retrieve data from the index library based on the search terms, compare the retrieved data with the test question resource data for duplicate detection, and provide feedback on duplicate test questions, non-duplicate test questions, and duplicate detection results to be manually confirmed.

4. An indexing device for test question retrieval, characterized in that: include: The test question judgment module obtains test question resources and judges test question attributes; An index building module builds an index based on the question attributes of the question resource using a preset text preprocessing strategy corresponding to the question attributes; The index building module builds an index based on the subject of the test question resource using a preset text preprocessing strategy of the corresponding subject, including: The test question judgment module judges that the test question resource is a science test question, and the index creation module performs word segmentation on the test question resource, extracts the formula in the test question resource, and creates an index separately with the formula as the index word; The extracting formulas from the test question resources and establishing a separate index using the formulas as index words includes: converting the formula format from the extracted test question resources into a unified set format, and then establishing a separate index using the formulas in the unified format as index words; The formulas in the extracted test question resources are in LaTeX format, which is converted into formula strings that can be recognized by OCR; The extracting formulas from the test question resources and establishing a separate index using the formulas as index words includes: automatically correcting easily identifiable errors in the formulas from the extracted test question resources, establishing a separate index using the corrected formulas as index words, and establishing a redundant index using the original formulas from the extracted test question resources as index words; When the test question is a math test question, the math symbols in the test question are retained, the math formulas are extracted, the math formulas are easy to identify and errors are automatically corrected, and the math formulas are indexed separately; When the test question is a physics test question, the physics symbols in the test question are retained, the physics formulas are extracted, the physics formulas are easy to identify and errors are automatically corrected, and the physics formulas are indexed separately; When the test question is a chemistry test question, the chemical symbols in the test question are retained, the chemical formula is extracted, the chemical formula is easy to identify and errors are automatically corrected, and the chemical formula is indexed separately; The index building module builds an index based on the subject of the test question resource using a preset text preprocessing strategy of the corresponding subject, including: The test question judgment module judges that the test question is a test question for infants and young children. The index establishment module performs word segmentation on the test question resource, extracts the test question images in the test question resource, and establishes an index based on the test question images. The index building module builds an index based on the subject of the test question resource using a preset text preprocessing strategy of the corresponding subject, including: The test question determination module determines that the test question is a test question of other liberal arts subjects, and the index establishment module deletes special characters in the test question resource, performs word segmentation according to Chinese semantics, and establishes an index according to the segmented items; The index establishment module establishes the index according to the preset text preprocessing strategy of the corresponding subject of the subject resource. The test question judgment module judges that the test question is an English test question, the index establishment module retains spaces, performs word segmentation according to the spaces, and establishes the index according to the items of the word segmentation; The index establishment module establishes the index according to the preset text preprocessing strategy of the corresponding subject of the subject resource. The test question judgment module judges that the test question is a reading test question, the index establishment module extracts the stem part of the test question resource after performing word segmentation on the test question resource, and performs compression fragmentation on the stem part, and establishes the index according to the compressed stem fragments; The subject symbols related to the subject of the test question are retained before the test question resource is segmented, and the subject symbols are normalized. 5.An electronic device, comprising a processor and a memory, the memory being configured to store a computer executable program, and wherein: When the computer executable program is executed by the processor, the processor performs an index establishment method for test question retrieval according to any one of claims 1-3.

Images