An efficient financial data retrieval method and a financial data processing system

Abstract

The present invention relates to the technical field of data processing, and specifically relates to a method for efficiently retrieving financial data and a financial data processing system. The method first analyzes the fuzzy retrieval behavior of the retriever in financial data to obtain the actual expected retrieval community of the retriever for financial data, and then respectively obtains the information effectiveness of different financial data in the expected retrieval community, and obtains the information effectiveness increment of different financial data in the expected retrieval community according to multiple fuzzy retrieval behaviors. Then, according to the effectiveness and effectiveness increment of the financial data, the final expected value of the financial data is obtained. Finally, according to the financial data obtained by the retriever in the last retrieval behavior under one retrieval purpose and the final expected value corresponding to each financial data, the retrieval result is displayed to the retriever. The present invention can give the retriever more accurate financial data that meets the retriever's expectations.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an efficient financial data retrieval method and a financial data processing system. Background Art

[0002] In the operation and management of modern enterprises, financial data is not only the basis for internal auditing, financial statement preparation and tax declaration of enterprises, but also an important information source for external investors, some institutions and other stakeholders to evaluate the health status and investment value of enterprises. Financial data covers all aspects of an enterprise's income, expenditure, assets, liabilities, cash flow, etc., and is almost involved in every link of the enterprise's operation. Therefore, accurate and efficient financial data processing is particularly important. Due to the large volume and complex content of enterprise financial data, and continuous updates with the changes of enterprise operations, how to quickly extract valuable information from the huge financial data has become an urgent problem to be solved.

[0003] The existing financial data retrieval methods generally involve the retriever setting corresponding keyword restrictions and then using a full-text search engine to conduct a global search of the financial database. This retrieval method can provide relatively effective retrieval results when the keyword input is very accurate. However, the common situation is that most of the retrieved keywords are fuzzy words, and the global search method is an independent retrieval process, that is, each retrieval is not connected in this retrieval algorithm. This retrieval usually returns a large number of documents, even completely irrelevant results, and this retrieval process is very inefficient, causing inconvenience to the staff in retrieving and reviewing financial data. Summary of the Invention

[0004] In order to solve the technical problem of the low efficiency in the process of financial data retrieval in the prior art, the purpose of the present invention is to provide an efficient financial data retrieval method and a financial data processing system, and the specific technical solutions adopted are as follows:

[0005] In the first aspect, an efficient financial data retrieval method is provided, and the method includes:

[0006] Step S1: Collect financial data, establish an index, and construct a financial database;

[0007] Step S2: Obtain the validity of a financial data according to the browsing behavior of a financial data in each retrieval behavior of the retriever under one retrieval purpose;

[0008] Step S3: Obtain the validity increment of a financial data according to the number of times a financial data appears in multiple retrieval results, the number of times it is browsed, and the similarity of the retrieval keywords each time it appears in multiple retrieval behaviors of the retriever under one retrieval purpose;

[0009] Step S4: Obtain the final expected value of the financial data based on the validity and validity increment of the financial data.

[0010] Step S5: Display the retrieval results to the retriever according to the financial data obtained from the last retrieval behavior of the retriever under one retrieval purpose and the final expected value corresponding to each financial data.

[0011] Further, the specific steps of Step S1 include:

[0012] Collect financial data, clear or correct incomplete and inaccurate financial data, and perform standardization processing on the financial data.

[0013] Establish an index for each piece of financial data, and construct a financial database based on the financial data after indexing.

[0014] Further, before Step S2 and after Step S1, it also includes:

[0015] Obtain the satisfaction degree of the retriever for the first retrieval result according to the operation behavior of the retriever on the first keyword retrieval result.

[0016] Obtain the possibility of the retriever to conduct a second retrieval according to the satisfaction degree of the retriever for the first retrieval result.

[0017] If the possibility of the retriever to conduct a second retrieval is greater than or equal to the preset possibility threshold, then continue with Step S2 - Step S5; if the possibility of the retriever to conduct a second retrieval is less than the preset possibility threshold, then there is no need to continue with Step S2 - Step S5.

[0018] Further, the specific method of obtaining the satisfaction degree of the retriever for the first retrieval result according to the operation behavior of the retriever on the first keyword retrieval result is: Obtain the satisfaction degree of the retriever for the first retrieval result according to the time interval between the first keyword retrieval and the second keyword retrieval of the retriever, the semantic similarity of the keywords between the first keyword retrieval and the second keyword retrieval, the average viewing duration of the financial data browsed by the retriever in the first retrieval result, the total number of citation operations on the financial data in the first retrieval result, and the total number of financial data browsed by the retriever in the first retrieval result.

[0019] Further, the time interval between the first keyword search and the second keyword search by the searcher, the average viewing duration of the financial data viewed by the searcher in the first search results, and the total number of reference operations on the financial data in the first search results are all positively correlated with the searcher's satisfaction with the first search results; the semantic similarity of the keywords in the first keyword search and the second keyword search, and the total number of financial data browsed by the searcher in the first search results are all negatively correlated with the searcher's satisfaction with the first search results.

[0020] Further, before step S2 and after step S1, it further includes: constructing a network diagram based on the financial data extracted according to the searcher's one-time search keyword, taking each financial data as a node, and dividing relevant communities as the search expectation community for the searcher's search based on the graph algorithm and the connectivity and density between nodes.

[0021] Further, in step S2, obtaining the validity of the financial data according to the browsing behavior of the searcher on a financial data in each search behavior under one search purpose of the searcher specifically includes: a search behavior of the searcher under one search purpose forms the search expectation community for this search. The searcher browses financial data in this search expectation community. When a financial data is browsed, the validity of this financial data is obtained according to the browsing duration of this financial data and the average duration of the other financial data browsed by the searcher. When a financial data is not browsed, the validity of this financial data is obtained according to the reciprocal of the total number of all un-browsed financial data.

[0022] Further, in step S3, the number of times a financial data appears in multiple search results and the similarity of the search keywords each time it appears in multiple search behaviors of the searcher under one search purpose are positively correlated with the incremental validity of this financial data, and the number of times a financial data is browsed in multiple search behaviors of the searcher under one search purpose is negatively correlated with the incremental validity of this financial data.

[0023] Further, in step S5, presenting the search results to the searcher according to the financial data obtained from the last search behavior of the searcher under one search purpose and the final expected value corresponding to each financial data specifically includes:

[0024] Screening the final expected values corresponding to all the financial data in the search expectation community obtained from the last search of the searcher, and screening out the financial data with a value less than the preset expected value threshold;

[0025] Sorting the remaining financial data in descending order according to the final expected values corresponding to these financial data from large to small, and the sorted remaining financial data is the display order of the relevant financial data corresponding to this search;

[0026] The search results are displayed in the order of display of the relevant financial data corresponding to this search.

[0027] In another aspect, the present invention provides a financial data processing system, comprising:

[0028] Financial database acquisition module, used to collect financial data, create indexes, and build a financial database;

[0029] A validity acquisition module is used to acquire the validity of financial data based on the browsing behavior of a financial data in each search behavior under a search purpose by the searcher;

[0030] The validity increment acquisition module is used to obtain the validity increment of a piece of financial data based on the number of times the financial data appears in multiple search results, the number of times it is browsed, and the similarity of the search keywords when it appears in multiple search behaviors under the same search purpose of the searcher;

[0031] A final expected value acquisition module, configured to acquire a final expected value of the financial data according to the validity and validity increment of the financial data;

[0032] The display result acquisition module is used to display the search results to the searcher based on the financial data obtained from the last search behavior under a search purpose of the searcher and the final expected value corresponding to each financial data.

[0033] The present invention has the following beneficial effects: by utilizing a dynamic incremental retrieval algorithm of reinforcement learning, the searcher's actual expected retrieval community for financial data is obtained by analyzing the searcher's fuzzy retrieval behavior in financial data, and then the information validity of different financial data in the expected retrieval community is obtained respectively, and dynamic learning adjustment of the information validity of different financial data in the expected retrieval community is performed based on multiple fuzzy retrieval behaviors, thereby more accurately realizing local joint learning incremental retrieval of financial data when the searcher uses fuzzy financial data retrieval, which can provide the searcher with more accurate financial data that meets the searcher's expectations. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 A flow chart of an efficient financial data retrieval method provided by one embodiment of the present invention.

[0036] Figure 2 A block diagram of a financial data processing system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0037] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of an efficient financial data retrieval method and financial data processing system proposed by the present invention. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention pertains.

[0038] The scenario targeted by the present invention is: when financial data is generally retrieved, it is necessary to search the data in the financial database. However, due to the large amount of data and the complexity of the content, obtaining accurate required information from the financial data often requires the assistance of extremely precise search formulas. However, when searching for financial data, ordinary personnel usually enter keywords multiple times for fuzzy search, resulting in the query usually returning a large number of documents, or even completely irrelevant results, making the overall search process very inefficient. In addition, because each search behavior of the conventional search method is usually an independent search behavior, the search keyword can only be manually adjusted according to the results of each fuzzy search to approach the final search desired key words. This independent search behavior cannot effectively learn the multiple fuzzy search behaviors together to adjust the search results.

[0039] The following describes in detail a specific scheme of an efficient financial data retrieval method and a financial data processing system provided by the present invention in conjunction with the accompanying drawings.

[0040] First, see Figure 1 , which shows a flow chart of an efficient financial data retrieval method provided by one embodiment of the present invention, the method comprising the following steps:

[0041] Step S1: Collect financial data, create indexes, and build a financial database.

[0042] Step S1 specifically includes: collecting financial data, clearing or correcting incomplete or inaccurate financial data, and standardizing the financial data; indexing each piece of financial data, and building a financial database based on the indexed financial data.

[0043] More specifically, first, collect the required financial data through methods such as the enterprise's internal ERP (Enterprise Resource Planning) system, external financial reports, and third-party APIs (Application Programming Interfaces), and clean the collected financial data to ensure the integrity and accuracy of the data; then, standardize these financial data, including unifying the units of relevant parameters in the financial data and the time formats in the financial data, etc.; finally, establish an index for each piece of financial data and create a database for the indexed financial data. Thus, the construction of the financial database is completed.

[0044] When retrieving the financial data in the financial data database, according to the information feedback of the retrieval, it is generally divided into two types of results: The first type of result is that the retrieved result is the accurate result required by the retriever, then at this time the retriever can obtain effective financial information and the retrieval ends; The second type of result is that the retrieved result is an inaccurate result, then at this time the retriever cannot obtain the required effective financial information and needs to conduct a secondary retrieval.

[0045] Generally speaking, the first type of retrieval result mentioned above is for precise searching. This retrieval mode requires the retriever to be very familiar with all the financial information in the financial database. However, because there is a large amount of financial data in the financial database, unless it is a special or professional person, this retrieval behavior is generally impossible. Most of the normal financial data retrieval behaviors belong to the second type of retrieval. In fact, when the retriever cannot obtain effective financial data information in the financial database, their behavior of conducting a second retrieval is a supplement and further limitation of the previous retrieval behavior, that is, the actual retrieval expectation of the retriever can be further obtained by analyzing the retrieval behavior of the retriever, so as to obtain the actual financial data information required by the retriever.

[0046] When conducting data retrieval, the above two types of results will occur, and the actual behaviors of the retrieval personnel for the two different results are different. Therefore, in this embodiment, by analyzing the retrieval behavior of the retriever and the usage behavior of the retriever for the retrieval result, the satisfaction degree of the retriever for the first retrieval result is obtained, so as to further obtain whether the current retrieval belongs to the first type of retrieval result or the second type of retrieval result. When the retrieval behavior belongs to the second retrieval result, the actual retrieval expectation of the retriever is obtained by further analyzing the connection of the qualifier in the retrieval process of the retriever, and is used to provide an efficient recommendation for the retrieval result.

[0047] Specifically, first, according to the operation behavior of the searcher on the first keyword search result, obtain the satisfaction degree of the searcher with respect to the first search result; then, according to the satisfaction degree of the searcher with respect to the first search result, obtain the possibility that the searcher conducts a second search; if the possibility that the searcher conducts a second search is less than the preset possibility threshold (the preset possibility threshold can be taken as 0.75), it is considered that the first search result belongs to the first type of search result, and under the current search purpose of the searcher, the accurate result required by the searcher has been obtained through the first search of the searcher, and this search ends; if the possibility that the searcher conducts a second search is greater than or equal to the preset possibility threshold, it is considered that the first search result belongs to the second type of search result, and it is necessary to further analyze the connection of the qualifiers in the search process of the searcher to obtain the actual search expectation of the searcher.

[0048] Among them, obtaining the satisfaction degree of the searcher with respect to the first search result according to the operation behavior of the searcher on the first keyword search result is specifically as follows: Obtain the satisfaction degree of the searcher with respect to the first search result according to the time interval between the first keyword search and the second keyword search of the searcher, the semantic similarity of the keywords in the first keyword search and the second keyword search, the average viewing duration of the financial data viewed by the searcher in the first search result, the total number of reference operations on the financial data in the first search result, and the total number of financial data browsed by the searcher in the first search result.

[0049] More specifically, when analyzing the satisfaction degree of the searcher with respect to the first search result, for the accurate information retrieved, the searcher generally has obvious click browsing or reference (transmission, download, etc.) of financial data, while for the inaccurate information retrieved, the searcher generally has unobvious clicks or browses multiple information in a short time, and there is no clear click browsing or reference to financial data. Therefore, based on this logic, the behavioral information of the searcher with respect to the first search result is quantified. Further, when the searcher conducts the first financial data search by inputting search keywords, if the search result is accurate, there is no need to conduct a second search, so the time interval between the two adjacent searches is relatively long, and there is a certain gap in the semantics of the keywords in the two searches; but if the search result is inaccurate, it is often necessary to conduct a second or multiple searches, so the interval time between the current search and the second search is relatively short, and the semantic similarity of the keywords is relatively high.

[0050] In this embodiment, the mathematical calculation formula for the satisfaction degree of the searcher with respect to the first search result is constructed as follows:

[0051]

[0052] In the formula, Indicates the time of the first keyword search by the searcher; Indicates the time of the second keyword search by the searcher; Indicates the semantic similarity of the keywords in the first keyword search and the second keyword search, ; Indicates the average duration for viewing financial data in the first search results; Indicates the total number of reference operations for all search results in the first search results; Indicates the total number of all the information browsed by the searcher in the first search results, ; Indicates the satisfaction degree of the searcher with the first search results.

[0053] In the above mathematical formula for calculating the satisfaction degree of the searcher with the first search results, Indicates the average time for the searcher to browse financial data. The shorter the average time, the more financial data the searcher browses in a shorter time, which means the lower the satisfaction degree of the searcher with the first search results; Indicates the proportion of reference operations in all financial data in the first search results; To avoid the denominator being zero, set the denominator in to , and obtain , which is positively correlated with the satisfaction degree of the searcher with the first search results; Indicates the time interval between the first keyword search and the second keyword search by the searcher. A longer time interval means the search results are accurate and there is no need for a second search. Therefore, is positively correlated with the satisfaction degree of the searcher with the first search results; The semantic similarity of the keywords in the first keyword search and the second keyword search , if there is a certain gap in the semantics of the keywords in the two searches, that is, is larger, it means the searcher is more dissatisfied with the first search results. Therefore, is negatively correlated with the satisfaction degree of the searcher with the first search results. In the above formula, is used as the benchmark value for the satisfaction degree of the searcher with the first search results, is used as the weight for the satisfaction degree of the searcher with the first search results.

[0054] Therefore, the time interval between the first keyword search and the second keyword search by the searcher, the average duration of viewing the financial data browsed in the first search results by the searcher, and the total number of reference operations on the financial data in the first search results are all positively correlated with the satisfaction of the searcher with the first search results; the semantic similarity of the keywords in the first keyword search and the second keyword search, and the total number of financial data browsed by the searcher in the first search results are all negatively correlated with the satisfaction of the searcher with the first search results.

[0055] It should be noted that the semantic similarity of the keywords in the first keyword search and the second keyword search can be obtained by using existing semantic analysis algorithms. For example, NLP (Natural Language Processing).

[0056] After obtaining the satisfaction of the searcher with the first search results, it is necessary to judge the possibility that the first search results belong to the second type of search results. In this embodiment, the mathematical calculation formula for the possibility that the constructed first search results belong to the second type of search results is as follows:

[0057]

[0058] In the formula, represents the satisfaction of the searcher with the first search results; represents the natural constant; represents the possibility that the first search results belong to the second type of search results.

[0059] In the above-mentioned mathematical calculation formula for the possibility that the first search results belong to the second type of search results, the higher the satisfaction of the searcher with the first search results, the lower the possibility of conducting the second search; the lower the satisfaction of the searcher with the first search results, the higher the possibility of needing to conduct the second search.

[0060] Preset possibility threshold , according to the empirical value, take , when , it is considered that the first search results belong to the first type of search results. Under the search purpose of this searcher, the accurate results required by the searcher have been obtained through the first search of the searcher, and this search ends; when , it is considered that the first search results belong to the second type of search results, and it is necessary to conduct the second search or even multiple searches to achieve the actual search expectation of the searcher.

[0061] In addition to determining the result that the retriever needs to conduct a second search, for the retriever's first search result and each subsequent search result, it is necessary to determine the community of the retriever's search expectation by using the retriever's search behavior, and then obtain the validity and validity increment of the financial data for different financial data in the community of the search expectation through non-initial search behaviors.

[0062] In this embodiment, a network diagram is constructed based on the financial data extracted according to the retriever's first search keyword. Each financial data is used as a node, and relevant communities are divided by using graph algorithms and the connectivity and density between nodes as the retriever's search expectation community for this search.

[0063] Specifically, taking the first search as an example, first, semantic analysis is performed on the keywords input by the retriever during the first search by using natural language processing (NLP) technology; then, financial data containing the keyword is extracted from the financial database, and a network diagram is constructed based on these financial data. Each financial data is used as a node, and the semantic similarity between each financial data and the keyword is used as the node distance. Relevant communities are automatically divided by using graph algorithms and according to the connectivity and density between nodes; finally, the automatically divided community is used as the retriever's search expectation community for the first search.

[0064] After determining that the retriever needs to conduct a second search and obtaining the retriever's search expectation community for the first search through the retriever's first search, it is necessary to further obtain the information validity of different financial data in the expected search community, and perform dynamic learning and adjustment of the information validity of different financial data in the expected search community according to multiple fuzzy search behaviors. Therefore, the following steps are further set in this embodiment.

[0065] Step S2: Obtain the validity of a financial data according to the browsing behavior of the financial data in each search behavior of the retriever under one search purpose.

[0066] Among them, obtaining the validity of a financial data according to the browsing behavior of the financial data in each search behavior of the retriever under one search purpose in step S2 specifically includes: a search behavior of the retriever under one search purpose forms the search expectation community for this search. The retriever browses financial data in this search expectation community. When a financial data is browsed, the validity of the financial data is obtained according to the browsing duration of the financial data and the average duration of the other financial data browsed by the retriever. When a financial data is not browsed, the validity of the financial data is obtained according to the reciprocal of the total number of all un-browsed financial data.

[0067] More specifically, first, when the first retrieval behavior occurs, according to the normal financial data retrieval process, there will be a display of the financial data results of the retrieval expected community related to the keywords in the first retrieval behavior. However, not all of the financial data in the expected community are what the retriever expects. Therefore, it is necessary to obtain the information validity for each piece of financial data in the expected community. The specific acquisition logic is as follows: Through the explicit and non-explicit behaviors of the retriever, the information validity of each piece of financial data is obtained.

[0068] The explicit behaviors of the retriever are divided into two categories: explicit affirmation and explicit rejection. Among them, the explicit affirmation behavior means that the retriever has obtained accurate information and does not apply the second type of retrieval results. Therefore, this embodiment mainly analyzes the explicit rejection behavior. The explicit rejection behavior refers to that for some financial data in the expected community, the retriever only has a simple browsing behavior (the shorter the browsing time, the lower the financial data validity), and there are no other reference operations. Then, these financial data are non-effective information for the retriever, so the information validity is low. Among them, the non-explicit behavior means that after the retrieval behavior occurs, there are some financial data in the expected community for which the retriever has no behavior, that is, neither browsing nor referencing, that is, the retriever has no corresponding information acquisition for these financial data and is in a vague cognitive state for these information. These financial data have higher validity compared to the explicitly rejected financial data.

[0069] In this embodiment, the mathematical calculation formula for the validity of each piece of financial data in the expected community constructed by the retriever is as follows:

[0070]

[0071] In the formula, represents the browsing duration of the retriever for the th piece of financial data in the expected community; represents the average browsing duration of the retriever for all pieces of financial data in the expected community; represents the total number of financial data that the retriever has not browsed among all pieces of financial data. Since there are generally financial data that the retriever has not browsed, is defined; represents the validity of the th piece of financial data in the expected community for the retriever.

[0072] In the above mathematical calculation formula for the validity of each piece of financial data in the expected community constructed by the retriever, When it is indicated that the nth financial data has been viewed by the searcher, the ratio of the viewing duration of the th financial data in the expected community by the searcher to the average viewing duration of all financial data in the expected community by the searcher is taken as the validity of the th financial data. Among them, setting serves to prevent the denominator from being 0; When it is indicated that the nth financial data has not been viewed by the searcher, an equal-weight operation is performed on all unviewed financial data (the reciprocal of the total number of unviewed financial data among all financial data) as the validity of the nth financial data.

[0073] It should be noted that the behavior of whether the above-mentioned nth financial data has been viewed by the searcher or not is a relative behavior. Therefore, and always have one item equal to 0. The validity of the th financial data in the expected community by the searcher can take the non-zero item.

[0074] Based on the above method, the validity of each financial data retrieved by the searcher in the expected community can be obtained according to the viewing behavior of each financial data in each retrieval behavior of the searcher under one retrieval purpose. In addition to analyzing the validity of each financial data retrieved by the searcher in the expected community, the associated characteristics under multiple retrieval behaviors of the searcher also need to be further considered to determine the final retrieval result display. Therefore, the following steps are further set in this embodiment.

[0075] Step S3: Obtain the validity increment of a financial data according to the number of times the financial data appears in multiple retrieval results, the number of times it is viewed, and the similarity of the retrieval keywords each time it appears in multiple retrieval behaviors of the searcher under one retrieval purpose.

[0076] Since the current retrieval behavior is a supplementary retrieval behavior on top of the first retrieval behavior, and a corresponding retrieval expectation community is generated after each keyword retrieval, and the current retrieval behavior is a supplement to the first retrieval behavior, there will be the same financial data in different retrieval expectation communities. In addition, when the retrieval results of each time are displayed, the data behavior of the retriever in different retrieval expectation communities determines the validity of the data information in the current community. Specifically, if in a certain retrieval behavior, the retriever simply looks up a certain financial data in the retrieval expectation community without citing the financial data, then even if this financial data is displayed in multiple retrieval results, it has low information validity for the retriever. However, if there are financial data that the retriever has not looked up when each retrieval result is displayed, the more times these financial data appear in the retrieval expectation communities of multiple retrieval results and the stronger their relevance to the retrieval keywords, the greater the likelihood that these financial data are the retrieval results expected by the retriever.

[0077] In this embodiment, the mathematical formula for the validity increment of the financial data obtained in multiple retrieval behaviors under one retrieval purpose is constructed as follows:

[0078]

[0079] In the formula, represents the total number of times the th financial data appears in all retrievals, where ; represents the total number of times the th financial data is viewed by the retriever in the times the result appears in all retrievals; represents the similarity of all the th financial data when it appears in all retrieval keywords; represents the validity increment corresponding to the th financial data in the th retrieval process.

[0080] In the mathematical formula for the validity increment of the financial data obtained in multiple retrieval behaviors under one retrieval purpose constructed above, since the current retrieval behavior is a supplementary retrieval behavior on top of the first retrieval behavior, the th financial data in the th retrieval process must be obtained in multiple retrieval behaviors under one retrieval purpose. In the formula, the th financial data in all The number of times the result appears in the Total number of times viewed by the searched person in With the th in the th search process, negatively correlated with the incremental validity corresponding to the th financial data. The fewer times the searched person views it, the greater the incremental information validity of the th financial data. In the above formula, the denominator takes to avoid the denominator being 0; the Total number of times the th financial data appears in all th searches in the th search process, positively correlated with the incremental validity corresponding to the th financial data. The more times the th financial data appears in all th searches, the greater the incremental information validity of the th financial data; further, the proportion of the total number of times the th financial data appears in all th searches in all searches, also positively correlated with the incremental validity corresponding to the th financial data in the th search process. The higher the proportion, the greater the incremental validity corresponding to the th financial data; further, the similarity of all th search keywords when the th financial data appears is also positively correlated with the incremental validity corresponding to the th financial data in the th search process. The higher the similarity of all th search keywords when the th financial data appears, the greater the incremental information validity of the th financial data.

[0081] Therefore, in multiple search behaviors of a searcher under one search purpose, the number of times a financial data appears in multiple search results, the similarity of search keywords each time it appears, are positively correlated with the incremental validity of the financial data, and the number of times a financial data is viewed by the searcher under one search purpose is negatively correlated with the incremental validity of the financial data.

[0082] Step S4: Obtain the final expected value of the financial data according to the validity and incremental validity of the financial data.

[0083] Specifically, by utilizing the effectiveness of the financial data in the expected community retrieved and the incremental effectiveness of such financial data under multiple retrieval actions, the final expectation value of the retriever for the financial data is further obtained.

[0084] In this embodiment, the mathematical calculation formula for the final expectation value of the financial data is constructed as follows:

[0085]

[0086] In the formula, represents the effectiveness of the th financial data in the expected community for the retriever; represents the incremental effectiveness corresponding to the th financial data in the th retrieval process; represents the final expectation value of the th financial data.

[0087] In the above constructed mathematical calculation formula for the final expectation value of the financial data, the effectiveness of the th financial data in the expected community for the retriever , and the incremental effectiveness corresponding to the th financial data in the th retrieval process are both positively correlated with the final expectation value

[0088] of the

[0089] th financial data. The greater the effectiveness of the financial data, the greater the incremental effectiveness of the financial data, and the greater the final expectation value of the financial data.

[0090] Step S5: Display the retrieval results to the retriever according to the financial data obtained from the last retrieval action of the retriever under one retrieval purpose and the final expectation value corresponding to each financial data.

[0089] Specifically, step S5 includes: screening the final expectation values corresponding to all the financial data in the retrieved expected community obtained by the retriever in the last retrieval, and removing the financial data with a final expectation value less than the preset expectation value threshold; sorting the remaining financial data in descending order according to the final expectation values corresponding to these financial data, and the sorted remaining financial data is the display order of the relevant financial data corresponding to this retrieval; displaying the retrieval results in sequence according to the display order of the relevant financial data corresponding to this retrieval.

[0090] More specifically, first, screen all the information in the retrieved expected community according to the final expectation value corresponding to each piece of information, and remove the information with too low final expectation value. The specific screening method is the threshold judgment method. For the final expectation value in the retrieved expected community that is less than or equal to the preset expectation value threshold The financial data is considered as irrelevant data for this search and will not be shown in the results. For data greater than the preset expected value threshold The financial data is considered as relevant data for this search and will be shown in the subsequent results. Then, for all the relevant data corresponding to this search, they are sorted in descending order according to the final expected value, and the sequence number of the sorted expected value is the display order of the relevant data corresponding to this search. Finally, the search results are shown in sequence according to the display order of the relevant data corresponding to this search.

[0091] In a second aspect, this embodiment provides a financial data processing system. Please refer to Figure 2 , which shows a block diagram of a financial data processing system provided by an embodiment of the present invention. The system includes:

[0092] A financial database acquisition module 101, configured to collect financial data, establish an index, and construct a financial database;

[0093] An effectiveness acquisition module 102, configured to obtain the effectiveness of a financial data according to the browsing behavior of the financial data in each search behavior of a searcher under a search purpose;

[0094] An effectiveness increment acquisition module 103, configured to obtain the effectiveness increment of a financial data according to the number of times a financial data appears in multiple search results, the number of times it is browsed, and the similarity of the search keywords each time it appears in multiple search behaviors of a searcher under a search purpose;

[0095] A final expected value acquisition module 104, configured to obtain the final expected value of the financial data according to the effectiveness and effectiveness increment of the financial data;

[0096] A display result acquisition module 105, configured to show the search results to the searcher according to the financial data obtained in the last search behavior of the searcher under a search purpose and the final expected value corresponding to each financial data.

[0097] Further, in the effectiveness acquisition module 102 of the system, obtaining the effectiveness of a financial data according to the browsing behavior of the financial data in each search behavior of a searcher under a search purpose specifically includes: a search behavior of a searcher under a search purpose forms a search expectation community for this search. The searcher browses financial data in this search expectation community. When a financial data is browsed, the effectiveness of the financial data is obtained according to the browsing duration of the financial data and the average duration of the other financial data browsed by the searcher. When a financial data is not browsed, the effectiveness of the financial data is obtained according to the reciprocal of the total number of all the un-browsed financial data.

[0098] An efficient financial data retrieval method and a financial data processing system provided by this embodiment. First, collect financial data, establish an index, and construct a financial database. Then, according to the browsing behavior of a financial data in each retrieval behavior of a retriever under a retrieval purpose, obtain the validity of this financial data. According to the number of times a financial data appears in multiple retrieval results, the number of times it is browsed, and the similarity of retrieval keywords each time it appears in multiple retrieval behaviors of a retriever under a retrieval purpose, obtain the validity increment of this financial data. According to the validity and validity increment of this financial data, obtain the final expected value of this financial data. Finally, according to the financial data obtained in the last retrieval behavior of a retriever under a retrieval purpose and the final expected value corresponding to each financial data, display the retrieval results to the retriever. This embodiment can give the retriever more accurate financial data that meets the retriever's expectations.

[0099] It should be noted that the above sequence of the embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0100] The above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features. And these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application and should all be included in the protection scope of the present application.

Claims

1. An efficient retrieval method for financial data, characterized in that: The method comprises: Step S1: Collect financial data, create indexes, and build a financial database; Step S2: obtaining the validity of the financial data based on the browsing behavior of the financial data in each search behavior under a search purpose by the searcher; Step S3: Obtaining the incremental effectiveness of the financial data based on the number of times a piece of financial data appears in multiple search results, the number of times it is browsed, and the similarity of the search keywords at each occurrence in multiple search behaviors under the same search purpose by the searcher; wherein the incremental effectiveness is obtained in the following manner: the number of times a piece of financial data appears in multiple search results and the similarity of the search keywords at each occurrence in multiple search behaviors under the same search purpose by the searcher are positively correlated with the incremental effectiveness of the financial data, and the number of times a piece of financial data is browsed in multiple search behaviors under the same search purpose by the searcher is negatively correlated with the incremental effectiveness of the financial data; Step S4: obtaining a final expected value of the financial data according to the validity and validity increment of the financial data; Step S5: Display the search results to the searcher based on the financial data obtained from the last search behavior under a search purpose of the searcher and the final expected value corresponding to each financial data.

2. The efficient financial data retrieval method according to claim 1, characterized in that: The step S1 specifically includes: Collect financial data, clear or correct incomplete or inaccurate financial data, and standardize financial data; Create an index for each piece of financial data, and build a financial database based on the indexed financial data.

3. The efficient financial data retrieval method according to claim 1, characterized in that: After step S1 and before step S2, the method further includes: According to the searcher's operation behavior on the first keyword search result, obtain the searcher's satisfaction with the first search result; According to the searcher's satisfaction with the first search result, the possibility of the searcher conducting a second search is obtained; If the possibility that the searcher conducts a second search is greater than or equal to the preset possibility threshold, then continue with steps S2 to S5; if the possibility that the searcher conducts a second search is less than the preset possibility threshold, then there is no need to continue with steps S2 to S5.

4. The efficient financial data retrieval method according to claim 3, characterized in that: The method of obtaining the searcher's satisfaction with the first search result based on the searcher's operational behavior on the first keyword search result is specifically as follows: obtaining the searcher's satisfaction with the first search result based on the time interval between the searcher's first keyword search and the second keyword search, the semantic similarity of the keywords in the first keyword search and the second keyword search, the average time the searcher spends viewing the financial data browsed in the first search result, the total number of reference operations on the financial data in the first search result, and the total number of financial data browsed by the searcher in the first search result.

5. The efficient financial data retrieval method according to claim 4, characterized in that: The time interval between the searcher's first keyword search and the second keyword search, the average time the searcher spends viewing the financial data browsed in the first search results, and the total number of reference operations on the financial data in the first search results are all positively correlated with the searcher's satisfaction with the first search results; the semantic similarity of the keywords in the first keyword search and the second keyword search, and the total number of financial data browsed by the searcher in the first search results are all negatively correlated with the searcher's satisfaction with the first search results.

6. The efficient financial data retrieval method according to claim 1, characterized in that: After step S1 and before step S2, the method also includes: constructing a network graph based on the financial data extracted from the search keyword of the searcher, taking each financial data as a node, and using the graph algorithm and the connectivity and density between the nodes to divide the relevant communities as the searcher's expected search communities for the search.

7. The efficient financial data retrieval method according to claim 6, characterized in that: In step S2, based on the browsing behavior of a financial data in each search behavior under a search purpose of the searcher, the validity of the financial data is obtained, specifically including: a search behavior under a search purpose of the searcher forms a search expectation community for the search, the searcher browses the financial data under the search expectation community, when a financial data is browsed, the validity of the financial data is obtained based on the browsing time of the financial data and the average time of the other financial data browsed by the searcher, when a financial data has not been browsed, the validity of the financial data is obtained based on the inverse of the total number of all the financial data that have not been browsed.

8. The efficient financial data retrieval method according to claim 6, characterized in that: In step S5, based on the financial data obtained from the last search behavior under a search purpose of the searcher and the final expected value corresponding to each financial data, the search results are displayed to the searcher, specifically including: Screening the final expected values corresponding to all financial data in the search expectation community obtained by the searcher in the last search, and filtering out financial data that is less than a preset expected value threshold; The remaining financial data are sorted in descending order according to the final expected values corresponding to these financial data. The remaining financial data after sorting will be the display order of the relevant financial data corresponding to this search; The search results are displayed in the order of display of the relevant financial data corresponding to this search.

9. A financial data processing system, characterized in that: The system comprises: Financial database acquisition module, used to collect financial data, create indexes, and build a financial database; A validity acquisition module is used to acquire the validity of financial data based on the browsing behavior of a financial data in each search behavior under a search purpose by the searcher; The validity increment acquisition module is used to obtain the validity increment of the financial data based on the number of times a piece of financial data appears in multiple search results, the number of times it is browsed, and the similarity of the search keywords at each occurrence in multiple search behaviors under a search purpose by the searcher; wherein the validity increment is obtained in the following manner: the number of times a piece of financial data appears in multiple search results and the similarity of the search keywords at each occurrence in multiple search behaviors under a search purpose by the searcher are positively correlated with the validity increment of the financial data, and the number of times a piece of financial data is browsed in multiple search behaviors under a search purpose by the searcher is negatively correlated with the validity increment of the financial data; A final expected value acquisition module, configured to acquire a final expected value of the financial data according to the validity and validity increment of the financial data; The display result acquisition module is used to display the search results to the searcher based on the financial data obtained from the last search behavior under a search purpose of the searcher and the final expected value corresponding to each financial data.

Images