Hot event prediction system based on Internet data collection

Abstract

The present invention relates to the field of Internet technology, and discloses a hot event prediction system based on Internet data collection, comprising a collection module for collecting main real-time attention data of various real-time events on a main network platform; a first comparison module, a screening module, a second comparison module, an integration module, and a prediction module. By comparing the real-time attention data of real-time events with historical hot event data of different public opinions, the similarity between the real-time attention data and the historical hot event data of each of the different public opinions is obtained, thereby predicting the development of the current event by finding the occurrence of known hot events corresponding to the historical hot event data with high similarity to the real-time attention data, thereby discovering and mastering public opinion information from the source, and effectively responding to the occurrence of hot events.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to a hot event prediction system based on Internet data collection. Background Art

[0002] With the increasing penetration of mobile internet, the scale of mobile internet users has expanded dramatically in recent years. The centers of online public opinion have shifted from traditional websites, forums, and blogs to Weibo, WeChat, and mobile news apps. The development of the mobile internet has transformed the landscape of online public opinion, with more and more netizens using their smartphones to express their views and attitudes on public affairs. Due to the distinct characteristics of the mobile internet compared to the fixed internet, mobile internet public opinion exhibits new characteristics across multiple elements of online public opinion. The most significant feature of mobile internet platforms is their ubiquity: ubiquity of subjects, time, and space. This ubiquity means that anyone, anywhere, at any time, can access the internet. The ubiquity of the mobile internet enables netizens to disseminate public opinion in real time. Events that would not likely spark public opinion on the fixed internet can become public opinion "touchpoints," and due to the immediacy of mobile internet dissemination, events can "ferment" more quickly. Although the carrier of public opinion on the mobile Internet is still mainly text, more and more netizens like to post pictures, voice and videos they take casually. Some netizens use the "long Weibo" tool to convert long texts into picture format and publish them, resulting in the fragmentation of online public opinion information. At the same time, the proportion of picture, video and audio public opinion information has increased, and the carrier of public opinion information has shifted from single text to rich media form.

[0003] While the emerging characteristics of mobile internet public opinion offer netizens faster, more diverse, and more authentic online interactions, they also present new challenges for public opinion management. A major challenge facing us in the new media era is how to effectively monitor and guide online public opinion, discover and grasp public opinion information at its source, and effectively respond to frequently occurring hot topics. Therefore, we propose a hot topic event prediction system based on internet data collection to predict online hot topics. Summary of the Invention

[0004] The purpose of the present invention is to provide a hot event prediction system based on Internet data collection to solve the following technical problems:

[0005] How to provide a prediction system that can accurately predict hot events.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] Hot event prediction system based on Internet data collection, including:

[0008] The collection module is used to collect the main real-time attention data of each real-time event of the main network platform;

[0009] Comparison module 1, for comparing the main real-time attention data with pre-stored historical hot event data of various different public opinions, and obtaining the main similarity between the main real-time attention data and the historical hot event data of various different public opinions;

[0010] A screening module is configured to receive the main similarities and sort the main similarities from large to small, compare the main similarity with the largest one in the sorting with a preset similarity threshold, and screen out real-time events corresponding to main similarities whose main similarities are not less than the preset similarity threshold;

[0011] Comparison module 2 is used to obtain secondary real-time attention data of the filtered real-time event on a secondary network platform other than the primary network platform, compare the secondary real-time attention data with pre-stored historical hot event data of each different public opinion, and obtain the secondary similarity between the secondary real-time attention data and the historical hot event data of each different public opinion;

[0012] An integration module, configured to integrate the primary similarity and the secondary similarity to generate a matching degree between the real-time event and each of the historical hot events of different public opinions;

[0013] The prediction module is used to judge and analyze the matching degree with a preset matching degree threshold and then issue an early warning signal.

[0014] Preferably, the process of comparing the main real-time attention data with the pre-stored historical hot event data of various public opinions is as follows:

[0015] Obtain the historical curve of the historical hot event data of each different public opinion over time based on the historical hot event data of each different public opinion;

[0016] Acquire a real-time curve of the main real-time attention data changing over time according to the main real-time attention data;

[0017] Obtain the real-time curve of the preset acquisition time period online, divide the historical curve into multiple historical sub-curves with the same preset acquisition time period, compare the real-time curve with the multiple historical sub-curves respectively, and obtain the curve length L that the real-time curve overlaps with each historical sub-curve 重合 , the number of overlapping pixels N 重合 and the sum of the area differences ΔS within a preset time period 总 .

[0018] Preferably, the primary similarity SIM:

[0019]

[0020] Among them, L实时 is the total length of the real-time curve, L 历史 is the total length of the historical sub-curve, N 历史 is the total number of pixels in the historical sub-curve, S 实时 is the total area of ​​the real-time curve within the preset acquisition time period, S 历史 is the total area of ​​the historical sub-curve within the preset acquisition time period, σ1, σ2, σ3 are preset weight coefficients, n is the number of historical sub-curves the historical curve is divided into, sim is the similarity between the real-time curve and the historical sub-curve within the preset acquisition time period, sim1, sim2, sim3...sim n They are the similarities between the real-time curve and each historical sub-curve within the preset collection time period.

[0021] Preferably, the process of integrating the primary similarity and the secondary similarity to generate the matching degree between the real-time event and each of the historical hot events of different public opinions is as follows:

[0022] Receiving the main similarities and sorting the main similarities from large to small, and selecting the top m main similarities;

[0023] receiving the sub-similarity and sorting the sub-similarity from largest to smallest, and selecting the top m sub-similarity;

[0024] The m primary similarities and m secondary similarities are integrated to obtain the matching degree.

[0025] Preferably, the matching degree Match:

[0026]

[0027] Among them, SIM i is the i-th primary similarity, SIM′ i is the i-th similarity, A and B are preset weight coefficients, i∈(1,m).

[0028] Preferably, the process of judging and analyzing the matching degree and the preset matching degree threshold and then issuing an early warning signal is:

[0029] If the matching degree Match is not less than the preset matching degree threshold Match0, an early warning signal is issued;

[0030] Otherwise, the real-time attention data of the real-time event in the next preset collection time period is collected.

[0031] Preferably, the prediction system also includes a warning module, which is used to mark the historical hot events corresponding to the top m main similarities and m secondary similarities according to the comparison module one, the comparison module two and the integration module, and use the historical development data of the marked historical hot events as a reference model of the event development trend of the real-time event for warning.

[0032] Preferably, the historical development data includes the event type, event time period, event dynamic development status, event public opinion trend and event development results of historical hot events.

[0033] Beneficial effects of the present invention:

[0034] This hot event prediction system based on Internet data collection compares the real-time attention data of real-time events with the historical hot event data of different public opinions to obtain the similarity between the real-time attention data and the historical hot event data of each of the different public opinions. It then predicts the development of current events by finding the known hot event occurrences corresponding to the historical hot event data with high similarity to the real-time attention data, thereby discovering and mastering public opinion information from the source, and effectively responding to the occurrence of hot events. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The present invention will be further described below with reference to the accompanying drawings.

[0036] Figure 1 This is a schematic diagram of the connection modules of the hot event prediction system of the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0038] See also Figure 1 As shown, the present invention is a hot event prediction system based on Internet data collection, comprising:

[0039] The collection module is used to collect the main real-time attention data of each real-time event of the main network platform;

[0040] Comparison module 1 is used to compare the main real-time attention data with the pre-stored historical hot event data of various public opinions, and obtain the main similarity between the main real-time attention data and the historical hot event data of various public opinions;

[0041] A screening module is used to receive the main similarities and sort the main similarities from large to small, and compare the main similarity with the largest one with a preset similarity threshold, and screen out the real-time events corresponding to the main similarities whose main similarity is not less than the preset similarity threshold;

[0042] The second comparison module is used to obtain the secondary real-time attention data of the selected real-time events on the secondary network platform outside the main network platform, compare the secondary real-time attention data with the pre-stored historical hot event data of each different public opinion, and obtain the secondary similarity between the secondary real-time attention data and the historical hot event data of each different public opinion;

[0043] The integration module is used to integrate the primary similarity and the secondary similarity to generate the matching degree between the real-time event and the historical hot events of different public opinions;

[0044] The prediction module is used to judge and analyze the matching degree with the preset matching degree threshold and then issue an early warning signal.

[0045] Through the above technical solution, real-time attention data of various real-time events on the network platform are collected, so as to analyze the development of the event through the change trend of the data. The real-time attention data includes but is not limited to the topic reading volume, discussion volume and forwarding volume of the event; then, by comparing the real-time attention data of the real-time event with the historical hot event data of different public opinions, the similarity between the real-time attention data and the historical hot event data of each of the different public opinions is obtained, so as to predict the development of the current event by finding the occurrence of known hot events corresponding to the historical hot event data with high similarity to the real-time attention data, thereby discovering and mastering public opinion information from the source, and then effectively responding to the occurrence of hot events;

[0046] The present invention first compares the main real-time attention data with the pre-stored historical hot event data of each different public opinion through the comparison module one, obtains the main similarity between the main real-time attention data and the historical hot event data of each different public opinion, then sorts the main similarities from large to small, compares the main similarity with the preset similarity threshold, and screens out the real-time events corresponding to the main similarity whose main similarity is not less than the preset similarity threshold. This process screens out the real-time events that may become hot events, and then obtains the secondary real-time attention data of the screened real-time events that may become hot events on the secondary network platform outside the main network platform through the comparison module two, compares the secondary real-time attention data with the pre-stored historical hot event data of each different public opinion, and obtains the secondary similarity between the secondary real-time attention data and the historical hot event data of each of the different public opinions, and then integrates the main similarity and the secondary similarity to generate the matching degree between the real-time event and the historical hot event of each different public opinion, and comprehensively considers the event development of the main network platform and the secondary network platform, thereby improving the accuracy of the hot event prediction system;

[0047] It should be noted that the collection and comparison processes of the primary and secondary network platforms are not in any particular order, and events are monitored simultaneously. The differences between the primary and secondary network platforms include but are not limited to mainstream news websites and new media websites.

[0048] The process of comparing the main real-time attention data with the pre-stored historical hot event data of various public opinions is as follows:

[0049] Obtain the historical curve of the historical hot event data of each different public opinion over time based on the historical hot event data of each different public opinion;

[0050] Obtain a real-time curve of the main real-time attention data changing over time according to the main real-time attention data;

[0051] Get the real-time curve of the preset acquisition time period online, divide the historical curve into multiple historical sub-curves with the same preset acquisition time period, compare the real-time curve with the multiple historical sub-curves respectively, and obtain the curve length L of the real-time curve and each historical sub-curve overlap 重合 , the number of overlapping pixels N 重合 and the sum of the area differences ΔS within a preset time period 总 .

[0052] Main similarity SIM:

[0053]

[0054]

[0055] Among them, L 实时is the total length of the real-time curve, L 历史 is the total length of the historical sub-curve, N 历史 is the total number of pixels in the historical sub-curve, S 实时 is the total area of ​​the real-time curve within the preset acquisition time period, S 历史 is the total area of ​​the historical sub-curve within the preset acquisition time period, σ1, σ2, σ3 are preset weight coefficients, n is the number of historical sub-curves the historical curve is divided into, sim is the similarity between the real-time curve and the historical sub-curve within the preset acquisition time period, sim1, sim2, sim3...sim n They are the similarities between the real-time curve and each historical sub-curve within the preset collection time period.

[0056] The above technical solution provides a method for calculating the main similarity, using the formula and Obtain, through the main similarity SIM, a comprehensive judgment is made on the similarity between the main real-time attention data and the historical hot event data, and then a comprehensive comparison is made between the similarity between the main real-time attention data and the historical hot event data, so as to predict the development of the current event based on the known hot event occurrence corresponding to the historical hot event data, thereby discovering and mastering public opinion information from the source, and then effectively responding to the occurrence of hot events;

[0057] It should be noted that the length of the overlap between the real-time curve and each historical sub-curve is L 重合 , the number of overlapping pixels N 重合 And the sum of the area differences within the preset acquisition time period ΔS 总 It is obtained by translating the curve up and down within the preset acquisition time period; the preset weight coefficients σ1, σ2, and σ3 are selectively set according to empirical data. When the real-time curve and the historical sub-curve trajectory match within the preset acquisition time period and more groups of alternating bands appear, there are multiple groups of area differences. At this time, σ3 accounts for a larger proportion than σ1 and σ2. When the real-time curve and the historical sub-curve trajectory have no overlapping pixels or very few overlapping pixels within the preset acquisition time period, σ1 accounts for a larger proportion than σ2 and σ3.

[0058] The process of integrating the primary similarity and the secondary similarity to generate the matching degree between the real-time event and the historical hot events of different public opinions is as follows:

[0059] Receive the main similarities and sort them from large to small, and select the top m main similarities;

[0060] Receive the sub-similarity and sort the sub-similarity from large to small, and select the top m sub-similarity;

[0061] The m primary similarities and m secondary similarities are integrated to obtain the matching degree.

[0062] Match:

[0063]

[0064] Among them, SIM i is the i-th primary similarity, SIM′ i is the i-th similarity, A and B are preset weight coefficients, i∈(1,m).

[0065] The above technical solution provides a matching degree calculation method, through the formula Obtain, through the matching degree Match, the event development of the primary network platform and the secondary network platform is comprehensively considered, thereby improving the accuracy of the hot event prediction system;

[0066] It should be noted that the preset weight coefficients A and B are selectively set according to empirical data, wherein the number of secondary network platforms and the corresponding number of secondary similarities SIM′ are greater than 1.

[0067] The process of judging and analyzing the matching degree and the preset matching degree threshold and then issuing an early warning signal is as follows:

[0068] If the matching degree Match is not less than the preset matching degree threshold Match0, an early warning signal is issued;

[0069] Otherwise, the real-time attention data of the real-time event in the next preset collection time period is collected.

[0070] The prediction system also includes a warning module, which is used to mark the historical hot events corresponding to the top m main similarities and m secondary similarities according to the comparison module one, the comparison module two and the integration module, and use the historical development data of the marked historical hot events as a reference model for the event development trend of real-time events for warning.

[0071] Historical development data includes the event type, event time period, event dynamic development status, event public opinion trend and event development results of historical hot events.

[0072] Through the above technical solution, it is possible to monitor, guide and intervene in real-time events that may become hot events in combination with the event type, event time period, event dynamic development status, event public opinion trend and event development results of historical hot events, so as to effectively respond to hot events; it should be noted that the historical development data can be queried on the website.

[0073] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A hot event prediction system based on Internet data collection, characterized by: include: The collection module is used to collect the main real-time attention data of each real-time event of the main network platform; Comparison module 1, for comparing the main real-time attention data with pre-stored historical hot event data of various different public opinions, and obtaining the main similarity between the main real-time attention data and the historical hot event data of various different public opinions; A screening module is configured to receive the main similarities and sort the main similarities from large to small, compare the main similarity with the largest one in the sorting with a preset similarity threshold, and screen out real-time events corresponding to main similarities whose main similarities are not less than the preset similarity threshold; Comparison module 2 is used to obtain secondary real-time attention data of the filtered real-time event on a secondary network platform other than the primary network platform, compare the secondary real-time attention data with pre-stored historical hot event data of each different public opinion, and obtain the secondary similarity between the secondary real-time attention data and the historical hot event data of each different public opinion; An integration module, configured to integrate the primary similarity and the secondary similarity to generate a matching degree between the real-time event and each of the historical hot events of different public opinions; A prediction module, configured to analyze the matching degree against a preset matching degree threshold and then issue an early warning signal; The process of comparing the main real-time attention data with the pre-stored historical hot event data of various public opinions is as follows: Obtain the historical curve of the historical hot event data of each different public opinion over time based on the historical hot event data of each different public opinion; Acquire a real-time curve of the main real-time attention data changing over time according to the main real-time attention data; Online acquisition of the real-time curve of the preset acquisition time period, dividing the historical curve into multiple historical sub-curves with the same preset acquisition time period, comparing the real-time curve with the multiple historical sub-curves, and obtaining the curve length of the real-time curve and each historical sub-curve. , the number of overlapping pixels and the sum of the area differences within a preset time period ; The main similarity SIM: ； ； in, is the total length of the real-time curve, The total length of The total number of pixels, is the total area of ​​the real-time curve within the preset acquisition time period, for The total area within the preset collection time period, 、 、 is the preset weight coefficient, is the number of segments of the historical sub-curves into which the historical curve is divided, is the similarity between the real-time curve and the historical sub-curve within the preset acquisition time period. 、 、 They are the similarities between the real-time curve and each historical sub-curve within the preset collection time period.

2. The hot event prediction system based on Internet data collection according to claim 1 is characterized in that: The process of integrating the primary similarity and the secondary similarity to generate the matching degree between the real-time event and each of the historical hot events of different public opinions is as follows: Receiving the main similarities and sorting the main similarities from large to small, and selecting the top m main similarities; receiving the sub-similarity and sorting the sub-similarity from largest to smallest, and selecting the top m sub-similarity; The m primary similarities and m secondary similarities are integrated to obtain the matching degree.

3. The hot event prediction system based on Internet data collection according to claim 2 is characterized in that: The matching degree : . ； in, is the i-th main similarity, is the ith similarity, and is the preset weight coefficient, i (1, ).

4. The hot event prediction system based on Internet data collection according to claim 3 is characterized in that: The process of judging and analyzing the matching degree and the preset matching degree threshold and then issuing an early warning signal is as follows: If the matching Not less than the preset matching threshold , an early warning signal is issued; Otherwise, the real-time attention data of the real-time event in the next preset collection time period is collected.

5. The hot event prediction system based on Internet data collection according to claim 4 is characterized in that: The prediction system also includes a warning module, which is used to mark the historical hot events corresponding to the top m main similarities and m secondary similarities according to the comparison module one, the comparison module two and the integration module, and use the historical development data of the marked historical hot events as a reference model of the event development trend of the real-time event for warning.

6. The hot event prediction system based on Internet data collection according to claim 5 is characterized in that: The historical development data includes the event type, event time period, event dynamic development status, event public opinion trend and event development results of historical hot events.

Images