32 results about "Concept tree" patented technology

The invention discloses an incremental concept calculation method for a college course system based on a concept tree. According to the method, firstly, the influence of newly added object nodes in a path tree on a concept is analyzed; secondly, according to the conversion relation of the path tree, the concept and a concept tree, the influence of newly added objects on the concept tree is obtained; finally, updated constraint conditions of the concept tree are given by analyzing the characteristics of the newly added objects, and concept tree updating is directly performed. The algorithm uses an object topology theory as the support, does not need actual operation object topology and has no hard requirement for a form background, and concept updating under dynamic data updating is achieved. By adopting the method, incremental form concept calculation is conducted on the form background composed by different semesters of learning courses and graduation requirement data of college students, different semesters of courses are gradually added into the composed concept tree, concept tree updating is performed to obtain the final updated concept tree, and accordingly knowledge systems of all the courses are obtained.

The invention provides a dielectric constant database retrieval method and system, including the following steps: (A) establishing a concept tree; (B) searching for user search terms in the concept tree to obtain more relevant search terms; (C) generating Query condition expression; (D) Retrieve the database; (E) Reorder the retrieval results. The present invention also provides a dielectric constant database retrieval system, which includes: database; database management module; concept tree generation module; interface input module; SQL statement generation module; database retrieval module; interface display module. The beneficial effects of the present invention are: the correlation degree of information retrieval in relational databases is improved, and the time for users to search for information is saved; the correlation degree is further quantified in combination with the concept tree, and a correlation degree parameter is introduced, so that when retrieving information for users, accuracy and fuzzy More options are provided; the relationship between search terms and concepts is used to dynamically calculate weights to further improve user experience.

The present invention relates to a virtual conference method and a three-dimensional virtual conference system oriented to semantic fidelity. The method includes: according to the semantic concept tree semantic structure; for each semantic concept of different levels in the semantic tree structure, setting a corresponding default Signal; extract the semantic description of the scene where the participants are in from the multimodal signal and update the private level semantic library; use the private level semantic library to update the public level semantic library; the server sends all the information in the public level semantic library to all clients The hierarchical semantic description composed of content, updates the corresponding private hierarchical semantic library on each client according to the hierarchical semantic description composed of all content; generates the virtual conference scene from the perspective of the participants; generates the virtual conference scene from the perspective of the participants during the meeting Virtual meeting scene. The method of the invention solves the problems of rough expressions and movements of virtual images, unreal meeting scenes, and difficult interaction between people and scenes, increases the scale of virtual meetings, and enhances the realism of meeting participants.

The invention provides a document semantic similarity calculation method. According to the method, the document retrieval workload is reduced, the working efficiency is improved, a threshold value setting method is adopted, and the similarity is calculated by intervals. The method specifically comprises steps as follows: A, constructing one or more sets of body libraries: each body library is formed by inputting of concept systems and main description terms, and in each body library, concepts form concept trees according to correlations and the concept trees form concept forests; B, calculating the semantic similarity: the semantic similarity between a query object vQuerym and a document vDocm is calculated with a tf-idf (term frequency-inverse document frequency) algorithm, the calculation formula is shown in the specification, tf is the number of times of appearance of the query object in the document, idf is the metric of the general importance of the query object, sim(cmi and cnj) is the semantic similarity between cmi and cnj, the calculation formula is shown in the specification, d is the distance between every two concepts in each concept tree, c is an automatic regulation parameter along with the system, p is the predefined correlation between cmi and cnj in each body, and the default is 1.

An emergency early warning grid division system and method based on spatial correlation analysis in the field of Internet disaster early warning technology, the system includes: an emergency event information integration module, an emergency event information filtering module and an emergency early warning grid division module. In this method, the geographical information of emergencies is used as the analysis clue of emergency early warning layout, and the expression mode of emergency spatial relationship information is firstly designed; then, the spatial relationship information of emergencies is extracted and integrated with the structural attribute information of emergencies Integrate and establish a constellation model of spatial facts; then filter out meaningless spatial relationship information according to the filtering algorithm based on frequency difference; finally, through the mining of hybrid spatial association rules, the spatial characteristics of the impacting events are obtained, and the composition covers different emergency warning levels emergency warning geographic grid. The present invention realizes the integration of the spatial relationship information and attribute information around the emergency, uses the concept tree to describe the spatial relationship information and event attribute information, and realizes the spatial fact constellation model that integrates the unstructured data and structured data of the emergency .

The application provides a text enhancement method, a text classification method and related devices. The method includes: obtaining the sentence content in the text corpus, and performing word segmentation on the sentence content to obtain the word after word segmentation; Similar words among the words whose similarity exceeds the threshold, and use similar words to randomly replace the sentence content to obtain multiple sentences; use multiple sentences to train the generation confrontation network, and obtain the generation confrontation network model; use the generation confrontation network model to generate extended sentences Sample; combine the expanded sentence sample with the sentence content in the text corpus to obtain an enhanced text data set. In the above implementation process, by using the generated adversarial network model obtained by training to generate extended sentence samples, since the generated adversarial network model learns the new change rules between similar words whose similarity exceeds the threshold in the concept tree, it can better to generate class-specific samples.

The invention relates to a data importing method of a clinical data integrating technology based on a translational medicine analysis platform, and belongs to the technical field of data integration. The method is used for importing of text type data and data in a source database, and comprises the following steps: preparing a data file, a column mapping file, a text replacing file and a file used for building a concept tree; building the concept tree, verifying whether data in the column mapping file meets a preset rule or not, reading and processing the data for the first time, building the concept tree again after adding markers of self-defined variables; importing a patient ID (identification), visiting ID data and patient information data into a platform database, carrying out related processing on data exception, network anomaly or platform database system exception. The method has the characteristics that the concept tree is built in a memory on the basis of tranSMART Batch, the self-defined variables are added, and the read-in of the data directly from the database is realized; an exception processing mechanism is added, so that the characteristics of high automation degree and strong adaptive capacity are realized.

The invention provides an image classification method, and in particular relates to an application method of concept cognition in the field of image classification. The present invention divides the image classification method based on the conceptual cognitive process into two modules: the cognitive training module, which is used to learn image concepts, and finally generates a concept tree to realize the image classification task; the image classification module, which uses the training module to obtain The concept tree of the image is classified; the method is described from the image feature description FD to the image form background K, and the necessary attribute set Base is generated from the image set form background K, and the coarse level of the concept tree CT is constructed by the necessary attributes. Background K Construct the finer levels of the concept tree CT, update the concept tree CT, obtain the optimal concept tree, use the optimal concept tree for classification, and solve formal concept cognition through image feature extraction technology, feature data processing technology and concept learning methods Difficult to apply to image problems.