A method for determining hotspot information, computer equipment and device
By extracting and combining shared keys from text elements, the method effectively addresses the dispersion of hot information due to varied descriptions, enhancing the accuracy and recall rate of hotspot identification.
Patent Information
- Application Number
- CN202011632878.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-31
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2040-12-31
AI Technical Summary
Existing methods struggle to accurately and efficiently determine hot information from large amounts of text due to dispersion caused by different descriptions of the same events, leading to low recall rates.
A method involving extracting elements from texts, determining shared keys for these elements, combining these keys to form combinations, and analyzing their frequency of occurrence to identify hotspot information.
This approach merges similar information with different descriptions, improving the recall rate of hot information determination and ensuring accurate identification without loss.
Smart Images

Figure CN112597772B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of information analysis, and in particular to a method, computer equipment and device for determining hotspot information. Background Art
[0002] With the continuous development of the Internet, the speed of information transmission is getting faster and faster. In a short period of time, a large amount of consulting news will emerge on the Internet, which is complicated and disordered. It is becoming more and more convenient for people to obtain information, but at the same time, it takes time to read carefully. It is difficult to obtain the current hot information from a large amount of text. In order to make the network services provided more immediate and timely, it is very important to accurately and efficiently determine the hot information in the current network. Summary of the invention
[0003] The main technical problem solved by the present application is to provide a hotspot information determination method, computer equipment and device, which can accurately and efficiently determine hotspot information.
[0004] In order to solve the above technical problems, a technical solution adopted in the present application is: to provide a method for determining hot spot information, including: obtaining a set of texts to be processed; extracting at least one element of each text in the set of texts to be processed; determining a shared key corresponding to each element; wherein each shared key corresponds to one element or multiple elements with the same semantics; combining different shared keys to obtain a shared key combination; and determining the hot spot information according to the frequency of occurrence of the shared key combination in the set of texts to be processed.
[0005] In order to solve the above technical problems, another technical solution adopted by the present application is: providing a computer device, including a processor, the processor is used to execute instructions to implement the above hot spot information determination method.
[0006] In order to solve the above technical problems, another technical solution adopted by the present application is: to provide a device with a storage function, storing program data, which can be read by a computer, and the program data can be executed by a processor to implement the above hotspot information determination method.
[0007] The beneficial effect of the present application is that, different from the prior art, the present application provides a method for determining hot information. The method determines the shared key of the elements and combines the shared keys to determine the hot information. The same content with different descriptions can be merged so that the popularity of the hot information will not be dispersed. The method can accurately determine the frequency of occurrence of events from a text set, improve the recall rate of hot information, and avoid the loss of hot information due to different descriptions. BRIEF DESCRIPTION OF THE DRAWINGS
[0008] Figure 1is a flowchart of a method for determining hotspot information according to an embodiment of the present application;
[0009] Figure 2 is a flowchart of a method for determining hotspot information according to another embodiment of the present application;
[0010] Figure 3 is a flowchart of a method for updating a synonymous element library according to an embodiment of the present application;
[0011] Figure 4 is a schematic diagram of a hotspot information push interface according to an embodiment of the present application;
[0012] Figure 5 is a structural diagram of a hotspot information determination system according to an embodiment of the present application;
[0013] Figure 6 is a schematic diagram of the structure of a computer device according to an embodiment of the present application;
[0014] Figure 7 It is a structural schematic diagram of a device with storage function according to an embodiment of the present application. DETAILED DESCRIPTION
[0015] In order to make the purpose, technical solution and effect of the present application clearer and more specific, the present application is further described in detail below with reference to the accompanying drawings and examples.
[0016] The method for determining hot information in the embodiment of the present application can extract similar information with different description methods from a large amount of text information, determine the hot information in the text information, and effectively improve the accuracy and recall rate of determining the hot information. The embodiment of the present application can be applied to various information push systems, for example, a web page push system, a news push system or an article push system. The application scenario described in the embodiment of the present application is to more clearly illustrate the technical solution of the embodiment of the present application, and does not constitute a limitation on the end solution provided by the embodiment of the present application. It is known to those of ordinary skill in the art that, without paying creative labor, the technical solution provided in the embodiment of the present application is also applicable to similar technical problems in other application scenarios. The method for determining hot information in the present application can be implemented in a terminal device, or it can be implemented by devices such as a network background server, a server cluster or a service site.
[0017] In different texts, the same event may be described in different ways, which may cause the popularity of hot information to be dispersed, making it difficult to find hot information, resulting in a low recall rate of hot information. To solve this technical problem, this application provides a method for determining hot information, which is described in detail below.
[0018] See also Figure 1, Figure 1 1 is a flow chart of a method for determining hotspot information according to an embodiment of the present application. The method for determining hotspot information can be implemented by a hotspot information determination system. It should be noted that if there are substantially the same results, this embodiment does not take Figure 1 The process sequence shown is limited. Figure 1 As shown, the method includes:
[0019] Step 110: Obtain a text set to be processed.
[0020] In one embodiment, the set of texts to be processed may refer to a collection of texts that appear within a preset time period collected by the hot information determination system. The hot information determination system can obtain the set of texts to be processed from the data source. The data source may include but is not limited to authors of various news websites, authors of comment websites, or authors of social networking platforms, etc. When the data source publishes the text, the hot information determination system will obtain the newly published text. The preset time period can be set according to user needs, for example, it can be one hour, one day, one week or one month, etc. The text can be a letter, a news article, a comment article, an article on a social networking platform (such as an article published on a microblog, a blog, a public account, or a circle of friends), or web page content, etc.
[0021] Step 130: extract at least one element of each text in the text set to be processed.
[0022] In one embodiment, an element may refer to a constituent element of an event. The element types may include: time, person, government agency name, general agency name, disease, group name, address, law or regulation, document, project or brand, etc. An element may also refer to medical, engineering, judicial, etc.
[0023] The hot information determination system may use any method to extract elements from the text. For example, it may use an element extraction model to extract elements from the text.
[0024] Step 150: Determine the shared key corresponding to each element.
[0025] Among them, each shared key corresponds to one element or multiple elements with the same semantics; in one embodiment, each element corresponds to a shared key, but a shared key will correspond to one or more synonymous elements. The correspondence between elements and shared keys is shown in Table 1. For example, "demobilized soldier" corresponds to a shared key "46", but the shared key "46" corresponds to multiple synonymous elements, such as "demobilized soldier", "demobilized soldier", "demobilized officer", "demobilized veteran" "synonymous elements, that is, elements with the same semantics can refer to different ways of saying the same meaning.
[0026] Table 1: Example table of correspondence between elements and shared keys
[0027]
[0028] The sharing key corresponding to the element may be already stored or newly created. The hot information determination system may retrieve the sharing key corresponding to the element from other systems, or search and determine the sharing key corresponding to the element from a stored database, or create a new sharing key when there is no stored sharing key for the element.
[0029] Step 170: Combine different shared keys to obtain a shared key combination.
[0030] In one embodiment, different numbers of shared keys may be randomly combined. Shared keys correspond to words and cannot accurately describe an event. Shared keys may be combined to form phrases or sentences, thereby accurately describing an event. A variety of events may be obtained by randomly combining shared keys. By combining shared keys, multiple elements corresponding to different shared keys may be combined. Thus, a shared key combination may correspond to an element combination set including multiple element combinations. Specifically, N different shared keys can be combined to obtain a shared key combination. For example, shared key 32 and shared key 46 can be combined to obtain the shared key combination [32&46], and the corresponding element combination set is ["Anhui AA Co., Ltd." & "Demobilized Military Personnel", "AA Company" & "Demobilized Military Personnel", "Anhui AA" & "Demobilized Military Personnel", "AA" & "Demobilized Military Personnel", "Anhui AA Co., Ltd." & "Demobilized Soldiers", "AA Company" & "Demobilized Soldiers", "Anhui AA" & "Demobilized Soldiers", "AA" & "Demobilized Soldiers", "Anhui AA Co., Ltd." & "Demobilized NCO", "AA Company" & "Demobilized NCO", "Anhui AA" & "Demobilized NCO", "AA" & "Demobilized NCO", "Anhui AA Co., Ltd." & "Demobilized Veterans", "AA Company" & "Demobilized Veterans", "Anhui AA" & "Demobilized Veterans", "AA" & "Demobilized Veterans", ].
[0031] Step 190: Determine hotspot information according to the frequency of occurrence of the shared key combination in the text set to be processed.
[0032] The frequency of occurrence of a shared key combination in the text set to be processed may refer to the number or quantity ratio of texts containing any element combination corresponding to the shared key combination. The element combination corresponding to the shared key combination whose frequency of occurrence meets the preset conditions may be used as hotspot information. Hotspot information may also refer to texts containing a shared key combination that meets the preset conditions.
[0033] By determining the shared key of the elements and combining the shared keys to determine the hot information, the same content with different descriptions can be merged so that the popularity of the hot information will not be dispersed. This method can accurately determine the frequency of occurrence of events from the text set, improve the recall rate of hot information, and avoid the loss of hot information due to different descriptions.
[0034] See also Figure 2 , Figure 2 FIG. 1 is a flow chart of a method for determining hotspot information according to another embodiment of the present application. The method for determining hotspot information can be implemented by a hotspot information determination system. It should be noted that if there is substantially the same result, this embodiment does not take Figure 2 The process sequence shown is limited. Figure 2 As shown, the method includes:
[0035] Step 210: Obtain a text set to be processed.
[0036] The text set to be processed includes multiple texts.
[0037] Step 220: Perform reading comprehension and sequence tagging on the text, and extract at least one element of each text.
[0038] Reading comprehension of a text can extract mutually related elements in the text, for example, elements such as the person complained against, the person complained against's unit, the person complained against's position, etc., which are subordinate to each other. Specifically, reading comprehension of a text includes: determining the number of answers to preset questions in the text and the starting position of each answer, and extracting the elements at the corresponding position in the text according to the starting position of each answer. Reading comprehension of a text can be achieved through a reading comprehension model. A reading comprehension model is a pre-trained machine learning model, and the training samples are texts with multiple element labels with subordinate relationships.
[0039] Sequence labeling of text is used to extract common elements, such as institution names, document names, laws and regulations, or project names. Specifically, the specific process of sequence labeling of text includes: classifying each character in the text. If multiple consecutive characters belong to the same category, the multiple consecutive characters can be considered as an element, and the element is extracted. Sequence labeling of text can be achieved through a sequence labeling model. The sequence labeling model is a pre-trained machine learning model, and the training samples are texts with element labels.
[0040] The training texts of the reading comprehension model and the sequence labeling model can be texts from various fields, such as the medical field, engineering field, judicial field, or education field, etc. By using text samples from various fields to train the reading comprehension model and the sequence labeling model, the reading comprehension model and the sequence labeling model can be applied to the element collection of articles in various fields.
[0041] Elements can include many different types, such as address, name, or time. Names can include institution names, person names, project names, etc. Addresses can include addresses of various levels, for example, Anhui Province, Hefei City, Yaohai District, etc. are all address elements.
[0042] In one embodiment, the elements are regularized to obtain regularized elements. After the elements in the text are collected, some types of elements can also be regularized to obtain regularized elements. Regularization can refer to the complete expansion of the elements to obtain the full name corresponding to the element, so as to facilitate the subsequent steps. For example, in the regularization of address elements, the address extracted from the element can be split into the administrative division name and the specific address name. For the administrative division name, an abbreviation mapping table can be constructed according to the map data, as shown in Table 1. The abbreviations of the administrative divisions are regularized into full names through the abbreviation mapping table. For example, the regularized elements corresponding to the elements such as Hefei, Hefei City, Hefei, Anhui Province, Hefei City, Anhui Province, and Hefei City, Anhui Province are all Hefei City, Anhui Province.
[0043] Table 2, regularization processing mapping example table
[0044]
[0045] Step 230: Determine the shared key corresponding to the element.
[0046] In one embodiment, a search is performed in a synonymous element set library to determine a shared key corresponding to an element, wherein the synonymous element library includes a correspondence between an element and a shared key. In the synonymous element library, a shared key may correspond to a synonymous element set, and the synonymous element set includes one or more elements with the same semantics. The synonymous element library can be used to obtain different expressions of the same semantics more quickly.
[0047] You can directly search and match in the synonymous feature library to determine the shared key corresponding to the feature; when the feature is not included in the synonymous feature library, you can update the synonymous feature library to determine the shared key corresponding to the feature. For more details on how to update the synonymous feature library, see Figure 3 ,as well as Figure 3 Related description.
[0048] In the initialization phase, the hot information determination system can call the saved initial synonymous element library. The initial synonymous element library can be an artificially constructed element library with a certain amount of synonymous samples. By constructing a synonymous element library, it is only necessary to search in the library to determine the shared key corresponding to the element, which can greatly reduce the time for repeated matching; at the same time, words with the same semantics can be classified into one category to provide data support for subsequent hot information determination.
[0049] Step 240: Build a text index library.
[0050] In one embodiment, the corresponding text cluster of each shared key in the text set to be processed is determined; wherein each text in the corresponding text cluster contains at least one element corresponding to the shared key. That is, a text index library is constructed, and the text index library contains the mapping relationship between the shared key and the text. The mapping relationship means that the text corresponding to the shared key contains at least one element under the shared key. Furthermore, the text can be numbered, and the text index library can only contain the corresponding relationship between the element shared key and the text number. The mapping relationship between the shared key and the text number is shown in Table 3.
[0051] Table 3, Example table of mapping relationship between shared keys and text numbers
[0052]
[0053] As shown in Table 3, "demobilized military personnel" appears in text No. 33, "demobilized soldiers" appears in text No. 156, and "demobilized veterans" appears in text No. 378; among them, the shared keys corresponding to "demobilized military personnel", "demobilized soldiers" and "demobilized veterans" are all 46, so the text number list corresponding to shared key 46 is [33, 156, 378], that is, the text cluster corresponding to shared key 46 includes texts with text numbers 33, 156 and 378.
[0054] By establishing a text index library, a corresponding relationship between shared keys and texts can be established, which facilitates the subsequent determination of hot information.
[0055] Step 250: Combine different shared keys to obtain a shared key combination.
[0056] The shared keys of any number are continuously combined. Further, the number of the shared keys to be combined is less than or equal to 5. If the number of the shared keys to be combined is greater than 5, the processing time of the subsequent steps will be significantly increased, and the combination is not very meaningful.
[0057] In one embodiment, the shared keys are classified and the shared keys of different categories are combined. For example, the categories of shared keys may include addresses, people, or time. When combining, the shared keys of different categories are combined respectively, and a shared key combination containing addresses, people, and time can be obtained. The phrases obtained by combining the same category are largely meaningless and cannot describe the event. By combining shared keys of different categories, the number of meaningless combinations can be reduced and the processing efficiency can be improved.
[0058] Step 260: Determine the confidence level of the shared key combination.
[0059] The confidence of a shared key combination refers to the probability of other shared keys appearing in the combination when a shared key appears in the combination. Search statistics can be performed in the text index library to obtain the confidence of the shared key combination. Take the shared key combination "A&B" as an example to specifically illustrate the statistical process: count the number of texts in the text number list corresponding to shared key A and the text number list corresponding to shared key B, and calculate the proportion of the number of texts in the text number list corresponding to shared key A, which is the confidence of the shared key combination "A&B". The above is only an exemplary description, and the calculation of confidence can also be done in any other calculation method. In one embodiment, for a shared key combination with a confidence greater than a first threshold, the frequency of occurrence of the shared key combination in the text set to be processed is counted. That is, subsequent steps are performed for a shared key combination with a confidence greater than the first threshold. The first threshold is a pre-set fixed value, or it can be an associated value set according to the characteristics of the shared key combination. For shared key combinations with low confidence, they are directly abandoned. Through the screening of confidence, some meaningless shared key combinations can be removed, thereby reducing the processing time of subsequent steps and improving processing efficiency.
[0060] Step 270: Count the occurrence frequency of the shared key combination in the text set to be processed.
[0061] In one embodiment, the number of mutually overlapping texts in the text clusters corresponding to the shared keys in the shared key combination is determined, and the frequency of occurrence of the shared key combination is determined based on the number of mutually overlapping texts. The number of mutually overlapping texts refers to the number of texts that simultaneously appear in the text clusters corresponding to different shared keys in the shared key combination. Alternatively, a search and statistics can be directly performed in the text index library to obtain the number of overlapping texts in the text number set corresponding to the shared keys for the combination.
[0062] Step 280: Determine hotspot information.
[0063] In one embodiment, the element combination corresponding to the shared key combination whose occurrence frequency meets the second preset condition is used as hot spot information. The second preset condition may mean that the occurrence frequency is greater than a second threshold; or that the occurrence frequency is in the top N items. The second threshold may be a pre-set fixed value, or an associated value set according to the characteristics of the shared key combination.
[0064] In addition, the second preset condition can also be configured according to demand, for example, for a shared key combination with certain specific characteristics, a frequency of occurrence higher than a smaller set value can be determined as hot spot information. Alternatively, certain shared key combinations with characteristics can be determined preferentially.
[0065] Step 290: Push the text corresponding to the hot information to the user.
[0066] The text containing hot information can be pushed to the user so that the user can see the hot information first. The hot information to be pushed can also be personalized according to the needs of the user.
[0067] In one embodiment, the sharing key combinations that serve as hotspot information are classified to obtain the category of each sharing key combination; the user demand category is obtained; and the text corresponding to the sharing key combination with the same user demand category is pushed to the user. The sharing key combinations can be classified according to features such as time, place, or task, and the text is pushed according to the user's demand category. In addition, text can be selectively pushed to users based on other rules, such as: push all, push by demand category, push by frequency, push by frequency and confidence, push by demand address, push by demand organization, or push by person. Taking push by demand address as an example, the hotspot information can be divided in detail according to the level of the demand address, and events in the lower-level jurisdiction belong to the upper-level jurisdiction, thereby pushing the text corresponding to the hotspot information in a hierarchical manner. For example, Figure 4 As shown, Figure 4 is a schematic diagram of a hotspot information push interface according to an embodiment of the present application. Figure 4 Road widening in Anhui Province is a hot topic. Click Anhui Province to see the road widening hot events in each city. Click each city to see the road widening hot events in each county. It also displays the number of texts containing road widening hot events corresponding to each level of address.
[0068] When a feature cannot be found in the synonymous feature database, the synonymous feature database needs to be updated. For specific update methods, see Figure 3 . Figure 3 1 is a flowchart of a synonymous element library updating method according to an embodiment of the present application. The synonymous element library updating method can be implemented by a hotspot information determination system. It should be noted that if there is substantially the same result, this embodiment does not take Figure 3 The process sequence shown is limited. Figure 3 As shown, the method includes:
[0069] Step 310: Perform a fuzzy search to determine a list of candidate synonymous elements.
[0070] When an element is not found in the synonymous element library, the element is a new element. It is necessary to perform a fuzzy search on the new element in the synonymous element library to determine a list of candidate synonymous elements. Fuzzy search means that there is a certain difference between the information allowed to be searched and the information being searched. This difference is the meaning of "fuzzy" in the search. For example, it can be an element that is partially the same as the element being searched. Perform a fuzzy search in the synonymous element library to obtain multiple elements that have a little similarity with the new element, which constitute a list of candidate synonymous elements.
[0071] Step 320: Determine the similarity between the element and the candidate synonymous elements in the candidate synonymous element list. Calculate the similarity between the element and each candidate synonymous element to obtain the similarity between the element and each candidate synonymous element. The similarity calculation can be any calculation method, such as cosine similarity or cross similarity. The process of determining the similarity can be implemented by a matching model. The matching model can be a machine learning model. The matching model uses the Blending model fusion method to fuse DSSM, ESIM, Siamese Network and Decomposable Attention together to make the model generalize better. The training samples of the matching model can be multiple groups of element combinations with different similarities.
[0072] Step 330: Determine whether there are synonymous elements in the synonymous element library.
[0073] In one embodiment, a candidate synonymous element whose similarity satisfies a first preset condition is determined as a synonymous element of the element. The element refers to a new element. The first preset condition may mean that the similarity is maximum and greater than a third threshold. The third threshold is a pre-set fixed value or an associated value set according to the element feature.
[0074] If the answer is yes, that is, there are synonymous elements, then proceed to step 340 ; if the answer is no, that is, there are no synonymous elements, then proceed to step 350 .
[0075] Step 340: Use the shared key corresponding to the synonymous element as the shared key of the element.
[0076] The shared key corresponding to the synonymous element is determined as the shared key corresponding to the element. The synonymous element is an existing element in the synonymous element library, and the shared key corresponding to the synonymous element in the synonymous element library is used as the shared key of the new element.
[0077] Step 350: Generate a new shared key, associate the element with the newly generated shared key, and add the newly generated shared key to the synonymous element library.
[0078] If the synonymous element of the element does not exist in the synonymous element library, a new shared key is generated. The element is associated with the newly generated shared key, that is, the newly generated shared key is the shared key corresponding to the element. The element refers to the new element. The newly generated shared key and the corresponding new element are stored in the synonymous element library.
[0079] Step 360: Add the element to the synonymous element library.
[0080] The element is the new element mentioned above, that is, the element that cannot be searched in the synonymous element library. The element and the corresponding relationship between the element and the shared key are added to the synonymous element library to complete the update of the synonymous element library.
[0081] This application also provides a hot spot information determination system, see Figure 5 . Figure 5 1 is a schematic diagram of a hotspot information determination system according to an embodiment of the present application. In this embodiment, the hotspot information determination system 500 includes an acquisition module 510 , an extraction module 520 , a shared key determination module 530 , a combination module 540 and a hotspot information determination module 550 .
[0082] The acquisition module 510 is used to acquire a text set to be processed.
[0083] The extraction module 520 is used to extract at least one element of each text in the text set to be processed. The extraction module 520 can also be used to input the texts into a pre-trained element extraction model to extract at least one element of each text; wherein the element extraction model includes a reading comprehension model and a sequence labeling model.
[0084] Determine the shared key module 530, which is used to determine the shared key corresponding to each element. Each shared key corresponds to one element or multiple elements with the same semantics. Determine the shared key module 530 can also be used to obtain a synonymous element library; wherein the synonymous element library includes the correspondence between elements and shared keys; determine the shared key corresponding to the element from the synonymous element set library. Determine the shared key module 530 can also be used to perform a fuzzy search to determine a candidate synonymous element list if the element does not exist in the synonymous element library; use a matching model to determine the similarity between the element and the candidate synonymous element in the candidate synonymous element list; determine the candidate synonymous element whose similarity meets the first preset condition as a synonymous element of the element, and use the shared key corresponding to the synonymous element as the shared key of the element; and add the element to the synonymous element library. Determine the shared key module 530 can also be used to generate a new shared key if the synonymous element of the element does not exist in the synonymous element library; associate the element with the newly generated shared key, and add the newly generated shared key and the element to the synonymous element library. The shared key determination module 530 may also be used to perform regularization processing on the elements to obtain regularized elements, and determine the shared keys corresponding to the regularized elements.
[0085] The combination module 540 is used to combine different shared keys to obtain a shared key combination, wherein the number of shared key items to be combined is less than or equal to 5.
[0086] The module for determining hotspot information 550 is used to determine hotspot information according to the frequency of occurrence of the shared key combination in the text set to be processed. The module for determining hotspot information 550 can also be used to count the frequency of occurrence of the shared key combination in the text set to be processed; the element combination corresponding to the shared key combination whose frequency of occurrence meets the second preset condition is used as hotspot information. The module for determining hotspot information 550 can also be used to determine the corresponding text cluster of each shared key in the text set to be processed; wherein each text in the corresponding text cluster contains at least one element corresponding to the shared key; determine the number of overlapping texts in the corresponding text cluster of the shared key in the shared key combination, and determine the frequency of occurrence of the shared key combination based on the number of overlapping texts. The module for determining hotspot information 550 can also be used to determine the confidence of the shared key combination; for the shared key combination whose confidence is greater than the first threshold, the frequency of occurrence of the shared key combination in the text set to be processed is counted.
[0087] The hot information determination system 500 also includes a push module (not shown) for classifying the sharing key combinations that serve as hot information and obtaining the category of each sharing key combination; obtaining the user demand category; and pushing the text corresponding to the sharing key combination that is the same as the user demand category to the user.
[0088] See also Figure 6 , Figure 6 6 is a schematic diagram of a computer device according to an embodiment of the present application. In this implementation, the computer device 600 includes a processor 610 .
[0089] The processor 610 may also be referred to as a CPU (Central Processing Unit). The processor 610 may be an integrated circuit chip having signal processing capabilities. The processor 610 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.
[0090] The computer device 600 may further include a memory (not shown) for storing instructions and data required for the processor 610 to run.
[0091] The processor 610 is used to execute instructions to implement the method provided by any embodiment of the above-mentioned adaptive filtering method and any non-conflicting combination.
[0092] See also Figure 7 , Figure 7 It is a structural schematic diagram of a device with a storage function according to an embodiment of the present application. The device 700 with a storage function in the embodiment of the present application stores instructions, and when the instructions are executed, the method provided by any embodiment of the adaptive filtering method of the present application and any non-conflicting combination is implemented. Among them, the instructions can form a program file and be stored in the above-mentioned device with a storage function in the form of a software product, so that a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) executes all or part of the steps of the various implementation methods of the present application. The aforementioned device 700 with a storage function includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, or a terminal device such as a computer, a server, a mobile phone, and a tablet.
[0093] The beneficial effects that may be brought about by the embodiments of the present application include but are not limited to: (1) The recall rate of hot information is high. Synonymous events with different descriptions can be grouped into one event by using the shared key of synonymous elements to avoid heat dispersion; (2) It has a wide range of applications. When training the model, training samples from multiple fields are used, so that the method of the present application can be applied to the determination of hot spots in multiple fields; (3) It improves the user experience. Users can view the hot information of the fields or regions they want to know according to their own needs. It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or any other possible beneficial effects.
[0094] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.
[0095] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0096] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.
Claims
1. A method for determining hotspot information, characterized in that: include: Get the text set to be processed; Extracting at least one element of each text in the to-be-processed text set; Determining a shared key corresponding to each of the elements; Each of the shared keys corresponds to one element or multiple elements with the same semantics; Combining different shared keys to obtain a shared key combination; Determine a text cluster corresponding to each of the shared keys in the text set to be processed; wherein each text in the corresponding text cluster contains at least one element corresponding to the shared key; Determining the amount of mutually overlapping text in text clusters corresponding to the shared keys in the shared key combination; Determining the frequency of occurrence of the shared key combination in the to-be-processed text set based on the amount of the mutually overlapping texts; The element combination corresponding to the shared key combination whose occurrence frequency meets the second preset condition is used as hot spot information.
2. The method for determining hotspot information according to claim 1, characterized in that: The determining the shared key corresponding to each of the elements comprises: Searching is performed from a synonymous element library to determine the shared key corresponding to the element; wherein the synonymous element library includes the corresponding relationship between the element and the shared key.
3. The method for determining hotspot information according to claim 2, characterized in that: The determining of the shared key corresponding to each of the elements further includes: If the element does not exist in the synonymous element library, a fuzzy search is performed to determine a candidate synonymous element list; Determining the similarity between the element and candidate synonymous elements in the candidate synonymous element list; Determine the candidate synonymous elements whose similarity satisfies the first preset condition as synonymous elements of the element, The shared key corresponding to the synonymous element is used as the shared key of the element; and the element is added to the synonymous element library.
4. The method for determining hotspot information according to claim 3, characterized in that: The determining of the shared key corresponding to each of the elements further includes: If the synonymous element of the element does not exist in the synonymous element library, generating a new shared key; The element is associated with the newly generated shared key, and the newly generated shared key and the element are added to the synonymous element library.
5. The method for determining hotspot information according to claim 1, characterized in that: The determining of the shared key corresponding to each of the elements further includes: Regularization is performed on the elements to obtain regularized elements, and the shared key corresponding to the regularized elements is determined.
6. The method for determining hotspot information according to claim 1, characterized in that: After combining the different shared keys to obtain a shared key combination, the method further includes: determining a confidence level of the shared key combination; For the shared key combination whose confidence is greater than a first threshold, the occurrence frequency of the shared key combination in the to-be-processed text set is counted.
7. The method for determining hotspot information according to claim 1, characterized in that: Also includes: Classifying the shared key combination as the hotspot information to obtain a category of each shared key combination; Obtain user demand categories; The text corresponding to the sharing key combination of the same category as the user's requirement is pushed to the user.
8. The method for determining hotspot information according to claim 1, characterized in that: The step of extracting at least one element of each text in the to-be-processed text set includes: Perform reading comprehension and sequence tagging on the text, and extract at least one element of each text.
9. A computer device, characterized in that: It includes a processor, and the processor is used to execute instructions to implement the hot spot information determination method as described in any one of claims 1-8.
10. A device with a storage function, characterized in that: Program data is stored and can be read by a computer, and the program data can be executed by a processor to implement the hotspot information determination method as described in any one of claims 1-8.
