59 results about "Plagiarism detection" patented technology

The invention discloses a text content duplicate removal method. Whether texts with duplicate judgment to be implemented are the same as texts in a text library or not is judged through comparison of file fingerprints, comparison of main body content fingerprints and comparison of paragraph fingerprints of the texts. The text content duplicate removal method is low in computation overhead, high in duplicate judgment rate and high in response speed, duplicate judgment on the texts with the same contents and different composing types can be accurately carried out, and duplicate judgment on a small number of texts with different contents can be accurately carried out. The text content duplicate removal method is wide in application scope and can be applied to library duplicate judgment uploading, web spider webpage processing, paper and test paper plagiarism detection and the like.

The invention discloses an electronic homework plagiarism preventing system and method based on paragraph plagiarism detection. The electronic homework plagiarism preventing system comprises an electronic homework submitting device, an electronic homework receiving device, a plagiarism detecting queue device, an electronic homework analyzing device, a plagiarism detecting device and an electronic homework storage device. The method includes: the electronic homework receiving devices receives electronic homework submitted through the electronic homework submitting device and adds the same into the plagiarism detecting queue device; the plagiarism detecting queue device detects plagiarism by using paragraph as the detecting unit and combining effective paragraph judging. The method using paragraph as the detecting unit and combining a queue mechanism for plagiarism detection has the advantages that system efficiency and stability are increased while the electronic homework polarizing multiple electronic homework can be judged, the attribution of the original homework can be judged, and plagiarism prevention can be achieved.

Plagiarism may be detected, as disclosed herein, utilizing a database that stores documents for one or more courses. The database may restrict sharing of content between documents. A feature extraction module may receive edits and timestamp the edits to the document. A writing pattern for a particular user or group of users may be discerned from the temporal data and the documents for the particular user or group of users. A feature vector may be generated that represents the writing pattern. A machine learning technique may be applied to the feature vector to determine whether or not a document is plagiarized.

A system, method, and computer-readable medium for detecting plagiarism in a set of constructed responses by accessing and pre-processing the set of constructed responses to facilitate the pairing and comparing of the constructed responses. The similarity value generated from the comparison of a pair of constructed responses serves as an indicator of possible plagiarism.

The invention relates to a video plagiarism detection method and system. According to the method and system, a multi-frame difference method is combined on the basis of a difference hash image recognition algorithm, key frames of a video to be detected can be accurately extracted, and the problems that a single traditional hash algorithm is low in image processing accuracy, long in algorithm running time and the like are solved. The multi-frame difference method adopted by the invention is improved on the basis of the traditional inter-frame difference method, and compared with the traditionalalgorithm for extracting the video key frame, and the defect that only the boundary part can be extracted and the relatively complete region cannot be extracted is overcome. According to the method,a multi-frame difference method and a difference hash algorithm are combined, and the effect of extracting the video key frames in the aspects of frame number, accuracy and algorithm running time is optimal.

The invention refers to a Chinese numeral anti-plagiarism detection comparison system and method. The system includes passage storing, dismantling, and searing and comparing mechanisms and evaluating report mechanisms. In use processes, a user first uploads comparison passages to a center server by the passage storing mechanism for data storage; the center server distributes uploaded passages to an operation host computer which dismantles the comparison passages into sentence groups through the passage dismantling mechanism and uploads obtained sentences to a search engine sentence by sentence; the searing and comparing mechanism performs searching so as to obtain webpage or passages similar to dismantled passage sentences and to download the webpage or passages to the operation host computer to perform full text comparison between compared passages and similar webpage or passages, wherein full text comparison results can mark and show similar parts between compared passages and similar webpage or passages and note sources of webpage and can be passed back to a center server so that users can observe the comparison results.

The invention discloses a software similarity detection method based on dynamic control flow graph sequence birthmark. The method comprises the following steps: firstly assembling a starting address of a basic block in the plug-in program record program execution process and a branch hopping address at the ending of the basic block under a dynamic plug-in platform DynamoRIO; and then analyzing a log file, constructing a program dynamic control flow graph, and endowing the weight; establishing a weight sequence birthmark set WSB, and serving the length ratio of the WSB as parameter to compute the similarity of each pair of programs. By adopting the dynamic plug-in analysis and extracting the feature of the software in operation, the problems that the source code is absent and the reverse disassembling is difficult in the software plagiarism detection can be avoided; only the basic block starting address and the branch hopping condition are recorded in the dynamic plug-in analysis, and the expenditure is less in comparison with the birthmark based on the dynamic data flow tracking and like technology; the influence by unrelated interference information in the dynamic operation can beresisted, and the program similarity can be detected even if the software encrypts by using an encryption shell.

The invention provides a thesis plagiarism detection method and system. The method comprises the following steps: recording materials by a comparison library; recording segmented words and corresponding word classes by a segmented word library; carrying out word segmentation by a word segmentation module; generating segmented word class characteristic values by a segmented word characteristic value generation module; determining segmented word free vector dimensions by a segmented word free vector dimension determination module; generating segmented word simplified vector dimensions by a segmented word simplified vector dimension generation module; generating segmented word characteristic vectors by a segmented word characteristic vector generation module; carrying out word segmentation on a to-be-authenticated document by a to-be-authenticated document word segmentation module so as to obtain a segmented word result; determining segmented word free vector dimensions by a to-be-authenticated document segmented word free vector dimension determination module; generating to-be-authenticated document segmented word simplified vector dimensions by a to-be-authenticated document segmented word simplified vector dimension generation module; generating to-be-authenticated document segmented word characteristic vectors by a to-be-authenticated document segmented word characteristic vector generation module; and carrying out similarity comparison.

Plagiarism may be detected, as disclosed herein, utilizing a database that stores documents for one or more courses. The database may restrict sharing of content between documents. A feature extraction module may receive edits and timestamp the edits to the document. A writing pattern for a particular user or group of users may be discerned from the temporal data and the documents for the particular user or group of users. A feature vector may be generated that represents the writing pattern. A machine learning technique may be applied to the feature vector to determine whether or not a document is plagiarized.

The invention discloses a video plagiarism detection method, and the method comprises the steps: obtaining at least one base library video and query video, carrying out interval frame extraction to acquire a plurality of base library images and a plurality of query images; and inputting the plurality of base library images and the plurality of query images into a convolutional neural network for feature extraction to acquire base library video frame features and query video frame features; obtaining the similarity between each query video frame feature and each base library video frame feature, and taking the base library video frames of which the similarity is higher than a first preset threshold as neighbor matching frames; classifying the neighbor matching frames according to the codingidentifiers to generate at least one base library video frame set; selecting the base library video corresponding to the at least one base library video frame set as a candidate video; and forming avideo pair by the query video and each candidate video, and searching a suspected plagiarism fragment in each matched video pair through a network flow algorithm. In addition, the invention also provides a video plagiarism detection device, equipment and a medium.

The invention discloses a query generation method for source retrieval based on machine learning in plagiarism detection, relates to the technical field of information retrieval, in particular to a query generation technology in an information retrieval technology, and solves the problems of dependency on expert experience and lack of continuous improvement capability in a method for performing query generation by adopting a heuristic-based method in a source retrieval technology of the prior art. The method comprises the steps of obtaining a group of alternative query sets defined in the specification by adopting n existing query generation methods for a suspicious document fragment sk; sorting all alternative queries in the set to obtain a sorting list; and taking first m queries of the sorting list as queries, defined in the specification, of the suspicious document fragment sk. According to the method, an inherent research thought for the query generation method in the technical field of existing source retrieval is overcome, and a characteristic that different source retrieval methods have different source retrieval performances on the same suspicious document fragment is fully utilized.

The invention relates to a memory object access sequence-based software dynamic birthmark and plagiarism detection method. According to the method, an original program and a comparison program are compared by taking a function internal data structure with a mapping relationship with input data at the level of a high-level language and an access process in function execution as a program feature set; according to dirty spot tracking of externally input data in a program dynamic running process, access of a program input to memory objects in dynamic execution is captured, and stack frame changes corresponding to the memory objects in the program execution process are analyzed; and finally according to a memory object access sequence, software birthmarks are constructed, and comparison among different program software birthmarks is performed. The method has the advantages that independently developed programs with similar functions can be effectively identified, so that the misjudgment rate is low; and plagiarism detection behaviors under most conditions can be detected, so that the judgment omission rate is low.

The present invention provides a software local plagiarism detection method based on the dynamic instruction dependency graph birthmark. The method comprises: 1) using dynamic instrumentation to perform instruction level monitoring on a to-be-analyzed program, and capturing an instruction trajectory of each function; 2) for a dynamic instruction trajectory recording each function, carrying out data dependency and control dependency analysis, and constructing a dynamic instruction dependency graph birthmark; 3) calculating the similarity between instruction dependency graph birthmarks, and implementing the measure of similarity between functions; 4) based on the given threshold, constructing a list of suspicious functions for each function in the plaintiff program; 5) extracting the staticfunction call graph of the program, and performing precise pairing of the suspicious functions under the guidance of the calling dependency; and 6) based on the calling dependency, assembling matchedfunction pairs to generate a plagiarism evidence map, and measuring the proportion of suspected plagiarism part. According to the method provided by the present invention, local plagiarism detection is implemented by constructing a function-level birthmark; and the concept of a plagiarism evidence map is proposed for the first time, and the effectiveness of the evidence can be greatly enhanced.

The invention provides a plagiarism source retrieval sorting model construction method and a plagiarism source retrieval sorting method. According to the plagiarism source retrieval sorting model construction method, training samples are utilized to train a predetermined sorting logic regression model through an order pair-based sorting learning manner on the basis of a degree of aggregation between each plagiarism source document of a reference document and the reference document until a value of a predetermined loss function is minimum, the predetermined loss function includes first and second sub-loss functions, the first sub-loss function represents a loss caused by sorting errors of order pairs formed on the basis of the plagiarism source documents and non-plagiarism source documentsof the reference document, and the second sub-loss function represents a loss caused by sorting errors of order pairs formed by plagiarism source documents with different degrees of aggregation. The plagiarism source retrieval sorting method utilizes the above obtained sorting model to resort retrieval results of suspicious documents. The above technology of the invention can more accurately sortthe source retrieval results of the suspicious documents in plagiarism detection.

The invention discloses a fast similarity detection and evidence generation for large-scale programs based on code mapping and lexical analysis. Two-layer similarity detection method is used to detectplagiarism and generate evidence for large-scale software samples. Firstly, the code mapping method is used to analyze the coarse-grained similarity of large-scale programs and search for suspected similar programs quickly. Lexical analysis is then used to fine-grained analyze suspicious similar programs, determine program similarity and generate similar code evidence. Through the above methods,we can quickly and accurately find the plagiarized code in large-scale samples, and provide corresponding evidence to support it.

The invention discloses a text plagiarism detection method and system. According to the method, the number and the length of extracted sentence fingerprints are reduced by deleting short sentences andadopting a truncated character fingerprint mode; the sentence fingerprints are extracted by deleting names, place names, organization names, time and other redundant information in the sentences, accurate detection of slightly-changed plagiarism contents is achieved, for example, the situation that the names, the place names, the organization names and other contents are changed can also be detected, and robustness is enhanced. Compared with a traditional text plagiarism method, the technical scheme provided by the invention greatly reduces the computation burden, improves the detection speed, is more suitable for quickly retrieving the same or similar place of the to-be-detected file and the original text with the copyright in mass (billion level) original texts, and outputs all the plagiarism texts and the corresponding plagiarism degree of the to-be-detected file.

The invention provides a Tibetan and Chinese cross-language paper plagiarism detection method and a system, and relates to the technical field of information processing. According to the method, the twin long-short-term memory network model is trained and optimized through large-scale Tibetan-Chinese sentence pair corpora; the Tibetan-Chinese cross-language similarity calculation model based on the twin long-short-term memory network obtained by training is good in accuracy; according to the Tibetan-Chinese cross-language similarity calculation model based on the twin long-short-term memory network, when sentence pair similarity is detected, any priori knowledge and manual intervention are not needed, the accuracy of a sentence pair similarity value detection result is guaranteed, and therefore the plagiarism detection accuracy with the sentence pair similarity value as the judgment basis is guaranteed.

The invention provides a cross-linguistic plagiarism detection method based on multiple features. The method comprises the steps of 1, corpus building; 2, translation feature building, wherein according to the europeanized phenomenon and the translation body problem which generally occur in translated articles, translation feature building is conducted, by means of feature selection, the featuresare cleaned and filtered to obtain the effective features, and noneffective features or the features with unapparent effects are filtered out; 3, feature selection, wherein the effective features areselected from the multiple features for classifier training, and then whether or not the cross-linguistic plagiarism problem exists in a certain article or multiple articles is classified; 4, based onplagiarism detection corresponding to the features, for Chinese features, accurate English feature corresponding is conducted, and according to the translation features and the structural features, plagiarism results are correspondingly filtered and generated, and through WordNet, final confirmation is conducted on the plagiarism results. By means of the method, the cross-linguistic plagiarism problem can be solved according to the multiple kinds of features mined from translation.

A system, method, and computer-readable medium for detecting plagiarism in a set of constructed responses by accessing and pre-processing the set of constructed responses to facilitate the pairing and comparing of the constructed responses. The similarity value generated from the comparison of a pair of constructed responses serves as an indicator of possible plagiarism.

The invention discloses a multi-language code plagiarism detection method based on a pseudo twin network, and the method comprises the steps: 1), obtaining basic data which comprises a pre-training data set and a multi-language code plagiarism detection training data set; 2) preprocessing the pre-training data set to obtain an accurate mark vector; 3) preprocessing the multi-language code plagiarism detection training data set to preliminarily judge whether the code is plagiarism or not; and 4) further judging whether the plagiarism exists in the multi-language code plagiarism detection training data set or not. According to the method, the limitation that code structure characteristics are not considered when codes are taken as texts to be processed in an existing multi-language code plagiarism detection method based on machine learning is broken through; in combination with structural characteristics of codes based on an abstract syntax tree, a convolutional neural network, a bidirectional long-short-term memory artificial neural network and a novel attention neural network are embedded into a pseudo twin network, so that multi-language code plagiarism detection is realized, andthe code plagiarism detection efficiency and precision are effectively improved.

A method for plagiarism detection of multithreaded program based on a thread slice birthmark includes steps of: 1) monitoring target programs during executing, real-time identifying system call, and recording related information comprising thread IDs, system call numbers, and return values; then pre-treating the information for obtaining a valid system call sequence Trace; 2) slicing the valid system call sequence Trace, for generating a series of thread slices Slice identified by the thread IDs; 3) generating dynamic thread slice birthmarks Birth of all the thread slices of two programs; 4) respectively generating corresponding software birthmarks PB1 and PB2 of the P1 and the P2 ; 5) matching based on a max bilateral diagram for calculating a max similarity between the software birthmarks PB1 and PB2; and 6) determines whether the program is plagiarized or not according to an average value of the birthmark similarity and a given threshold ε.

The invention relates to an image design work plagiarism detection method based on an adversarial network. Existing image design works are lack of tampered samples. The problem that traditional deep neural network training is difficult to realize is solved; design of brand-new logic strategy, tampering the mask image for a model generated by the generation network according to the original plagiarism image; an artificially marked tampered mask image is combined; judging which mask image is manually marked by using a judgment network; therefore, the accuracy of the generation network and the judgment network is represented based on whether the judgment result is correct or not; executing a corresponding feedback training operation; the accuracy of the two networks is continuously improved;in this way, through continuous confrontation iteration, the accuracy of the two networks reaches a final balance state, the obtained generation network is an image detection model and has excellent work plagiarism recognition performance, a tampered mask image of a to-be-discriminated image can be obtained by applying the image detection model, and tampering detection of an image design work is efficiently achieved.

The invention discloses a code plagiarism detection method and system based on a program language teaching practice platform, and the method comprises the steps: obtaining two job codes, carrying out the matching comparison based on the contents of the job codes, and determining the similarity of the two job codes; processing the similarity, and obtaining a final code plagiarism detection result of the two homework codes, wherein the step of processing the similarity comprises the step that a first parameter acts on the similarity data, and the first parameter is generated based on editing operation characteristics of a student when the student edits the homework codes. According to the method, the similarity of code texts is further integrated in combination with a specific language teaching practice use scene and editing operation characteristics of students during homework code editing, so that a homework code plagiarism result combined with a teaching scene is more accurate.

The invention relates to a test program plagiarism detection method based on a support vector machine. The method comprises: performing cutting and static analysis on a to-be-tested program and a testprogram to obtain a to-be-tested method mapping set and a test method mapping set; secondly, traversing the players pairwise, calculating the similarity of the test fragments, and summarizing to obtain a similarity set; secondly, selecting an appropriate kernel function and a reference point to establish a support vector machine model, and optimizing the support vector machine model; and finally,for other test programs, calculating a similarity set and inputting the similarity set into the support vector machine to judge plagiarism conditions among the test programs. The invention aims to fill the blank of a test program code similarity detection technology. The code plagiarism detection accuracy and precision of the test program are improved, so that developers are helped to test code plagiarism behaviors of competitors for competition automation detection, the manual detection link is omitted, the labor cost and the time cost are saved, and it is guaranteed that the competition ismore fair and justice.

The embodiment of the invention provides a multi-thread program plagiarism detection method based on dynamic birthmarks and related equipment. The method comprises the steps of inserting a custom function into a to-be-tested program by adopting a dynamic instrumentation technology to obtain a system call sequence; processing the system call sequence by utilizing a D-Kgram algorithm with a variableK value, and respectively generating a plurality of sub-sequences of which the gram lengths are different K values; performing single-thread screening on the plurality of sub-sequences to obtain a feature sub-sequence set; respectively constructing dynamic birthmarks of the original program and the suspicious program; converting the dynamic birthmarks into vectors, and obtaining the similarity between the original program and the suspicious program by using a cosine similarity method; and calculating the mean value of the similarity under multiple inputs, and obtaining a conclusion whether the suspicious program plagiarizes the original program or not according to the detection threshold. According to the method and the related equipment provided by the invention, the influence of threadinterleaving characteristics on the dynamic birthmarks can be effectively avoided, so that the plagiarism detection effect is better.

The invention relates to a test program plagiarism detection method based on test code fragment similarity. The test program plagiarism detection method comprises the following steps: for each to-be-tested method in a to-be-tested program, firstly, calculating a unique method identifier based on a class name, a method name and a parameter sequence; secondly, extracting all test code fragment setsfrom the test program, wherein each test fragment corresponds to one to-be-tested method; then, analyzing the similarity between the test fragments to obtain a similarity analysis report, and calculating a similarity value between the fragments; and finally, calculating the overall similarity degree value of the test programs by utilizing the similarity value of the test fragments, and judging theplagiarism condition between the test programs more accurately by utilizing the overall similarity degree value of the test programs. The test program plagiarism detection method aims to fill the blank of a test code similarity detection technology, and solves the problems of low precision of test code similarity analysis and low efficiency of test code plagiarism detection mainly depending on manual operation at present, thereby improving the efficiency and precision of test code similarity detection.

The invention provides a thesis plagiarism detection method and system. The method comprises the following steps: recording materials by a comparison library; recording segmented words and corresponding word classes by a segmented word library; carrying out word segmentation by a word segmentation module; generating segmented word class characteristic values by a segmented word characteristic value generation module; determining segmented word free vector dimensions by a segmented word free vector dimension determination module; generating segmented word simplified vector dimensions by a segmented word simplified vector dimension generation module; generating segmented word characteristic vectors by a segmented word characteristic vector generation module; carrying out word segmentation on a to-be-authenticated document by a to-be-authenticated document word segmentation module so as to obtain a segmented word result; determining segmented word free vector dimensions by a to-be-authenticated document segmented word free vector dimension determination module; generating to-be-authenticated document segmented word simplified vector dimensions by a to-be-authenticated document segmented word simplified vector dimension generation module; generating to-be-authenticated document segmented word characteristic vectors by a to-be-authenticated document segmented word characteristic vector generation module; and carrying out similarity comparison.

A method for plagiarism detection of multithreaded program based on a thread slice birthmark includes steps of: 1) monitoring target programs during executing, real-time identifying system call, and recording related information comprising thread IDs, system call numbers, and return values; then pre-treating the information for obtaining a valid system call sequence Trace; 2) slicing the valid system call sequence Trace, for generating a series of thread slices Slice identified by the thread IDs; 3) generating dynamic thread slice birthmarks Birth of all the thread slices of two programs; 4) respectively generating corresponding software birthmarks PB1 and PB2 of the P1 and the P2; 5) matching based on a max bilateral diagram for calculating a max similarity between the software birthmarks PB1 and PB2; and 6) determines whether the program is plagiarized or not according to an average value of the birthmark similarity and a given threshold ε.

The invention provides a Tibetan language paper plagiarism detection method and system, and relates to the technical field of modern education. Aiming at three different plagiarism phenomena of continuous text plagiarism, semantic rewriting plagiarism and translation plagiarism, the invention provides a longest common subsequence algorithm and an improved twin long-short-term memory network methodrespectively. Academic paper pre-detection based on abstract document vectors and a weight distribution strategy based on chapter positions are adopted to improve retrieval efficiency.

The invention discloses a homework duplicate checking method based on deep learning, and the method comprises the steps: obtaining student curriculum homework data and a homework template file, judging the format of a homework template, carrying out the question-cutting processing of the obtained homework, judging whether the question in the homework is a subjective question or an objective question, carrying out the text preprocessing of the answer of the subjective question in the homework after the question-cutting, employing the deep learning technology (namely a convolutional neural network model) to calculate the similarity between student homework, analyzing a similarity calculation result, clustering the student homework with high similarity, and generating a similarity report. In order to facilitate a teacher to check similar content conditions, similar contents among similar homework are marked. According to the method, text contents with similar job semantics can be found out, and the problem that many plagiarism detection methods are poor in anti-interference effect is solved.