Multi-factor comprehensive sorting method, system and equipment

Abstract

The invention belongs to the field of information retrieval, particularly relates to a multi-factor comprehensive sorting method, system and equipment, and aims to solve the problem of inaccurate weight judgment of each factor under different scenes and different user groups. According to the method, the influence of various factor types on the document can be accurately determined by extracting the factor types and performing priority ranking. According to the method and the device, the contribution degree of each factor to the comprehensive sorting of the documents is accurately measured through a quantification mode of endowing grading weights for factor types, so that the judgment on the weights of the factors under different scenes and different user groups is avoided. Retrieval standardization is carried out on the factor sequence formed by the factor types, and the accuracy and consistency of retrieval can be improved. Retrieval errors caused by inconsistent or nonstandard data are reduced. And comprehensively sorting the documents according to the grading weight sequence and the standard factor sequence. The efficiency and the accuracy of information retrieval can be improved, and a user can find required related documents more quickly.

Description

Technical Field

[0001] The present invention belongs to the field of information retrieval, and in particular relates to a multi-factor comprehensive ranking method, system and equipment. Background Art

[0002] In the field of information retrieval, traditional single-factor ranking methods have long dominated. Single-factor ranking refers to ranking search results based solely on a single factor. For example, in early search services, ranking was based solely on the degree of match with the search keyword. This method has the advantages of being simple and straightforward, easy to implement, and relatively low computational cost. However, since it only considers a single factor, it cannot fully reflect the true relevance of search results to user needs, making it difficult to meet users' diverse and complex search needs, significantly compromising the quality of search results and user experience.

[0003] With the rapid development of internet services, users are increasingly demanding search results, prompting the emergence of multi-factor comprehensive ranking. Multi-factor comprehensive ranking fully considers multiple factors across multiple dimensions, significantly improving the user search experience and becoming widely adopted in core business scenarios such as internet search and recommendations.

[0004] Although multi-factor comprehensive ranking has many advantages, in the field of academic document retrieval, how to scientifically and reasonably determine the weight of each factor is a major problem. Different factors have very different degrees of influence on the relevance of search results. Taking academic document search as an example, the authority of the document (such as whether it comes from a well-known academic journal) is compared with the keyword matching. In some professional search needs, the authority factor weight may be higher; while in ordinary knowledge popular science document searches, keyword matching may be more critical. However, in actual operation, it is difficult to accurately judge the exact weight of each factor in different scenarios and different user groups. Once the weight setting is inaccurate, it may lead to incorrect evaluation of the document, which in turn affects the accuracy and rationality of the sorting results. Summary of the Invention

[0005] In order to solve the above problems in the prior art, the present invention provides a multi-factor comprehensive ranking method, comprising:

[0006] Determine multiple documents to be returned based on a document query;

[0007] In a query-document pair set corresponding to a document query, extracting multiple factor types from the query-document pair set and prioritizing the multiple factor types according to a preset sorting method to obtain a factor type sequence corresponding to the query-document pair set; wherein the query-document pair includes a document query and a document, and a corresponding relationship between the document query and the document;

[0008] Assigning a hierarchical weight to each factor type in the factor type sequence to obtain a hierarchical weight sequence corresponding to the factor type sequence;

[0009] In the plurality of factor type sequences, a plurality of factor sequences are formed based on the same factor type, and each of the factor sequences is searched and normalized to obtain a normalized factor sequence corresponding to the factor sequence;

[0010] The returned multiple documents are comprehensively sorted according to the hierarchical weight sequence and the normalization factor sequence.

[0011] In an optional embodiment, the factor sequence is subjected to retrieval normalization, including:

[0012] Generate the full range of the factor sequence according to the element value of each factor type in the factor sequence;

[0013] The factor sequence is searched and normalized based on the full range.

[0014] In an optional embodiment, the performing search and normalization on the factor sequence based on the full range includes:

[0015] generating a normalized coefficient based on the full range;

[0016] Determine the product of the element value of each factor type in the factor sequence and the normalization coefficient to obtain a normalized sequence;

[0017] Each element value in the canonical sequence is rounded down to an integer to obtain a canonical factor sequence corresponding to the factor sequence.

[0018] In an optional embodiment, the comprehensively sorting the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence includes:

[0019] generating a scoring sequence according to the hierarchical weight sequence and the normalization factor sequence;

[0020] The plurality of documents are sorted in descending order based on the score sequence.

[0021] In an optional embodiment, generating a scoring sequence according to the hierarchical weight sequence and the normalization factor sequence includes:

[0022] The hierarchical weight sequence and the normalization factor sequence are matrix multiplied to obtain a scoring sequence.

[0023] In an optional embodiment, the weight coefficients assigned to the hierarchical weights meet the following proportional requirements:

[0024]

[0025] Among them, HW1, HW2, HW3, HW4, HW m-1 and HW m are the weight coefficients corresponding to the 1st, 2nd, 3rd, 4th, m-1 and mth factor types in the factor type sequence, and H is the grading coefficient.

[0026] In an optional embodiment, the factor types include at least: relevance type, authority type and timeliness type.

[0027] Another aspect of the present invention provides a multi-factor comprehensive ranking system, comprising:

[0028] An acquisition module for determining multiple documents returned based on a document query;

[0029] a priority module configured to extract, from a query-document pair set corresponding to a document query, a plurality of factor types and prioritize the plurality of factor types according to a preset sorting method to obtain a factor type sequence corresponding to the query-document pair set; wherein the query-document pair comprises a document query and a document, and a corresponding relationship between the document query and the document;

[0030] A weight module, configured to assign a hierarchical weight to each factor type in the factor type sequence to obtain a hierarchical weight sequence corresponding to the factor type sequence;

[0031] a processing module, configured to construct multiple factor sequences based on the same factor type from the multiple factor type sequences, and perform search and normalization on each of the factor sequences to obtain a normalized factor sequence corresponding to the factor sequence;

[0032] The sorting module is used to comprehensively sort the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence.

[0033] A third aspect of the present invention provides an electronic device, comprising:

[0034] at least one processor;

[0035] and a memory communicatively coupled to at least one of said processors;

[0036] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the above-mentioned multi-factor comprehensive ranking method.

[0037] In a fourth aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the above-mentioned multi-factor comprehensive ranking method.

[0038] The embodiments of the present application provide a multi-factor comprehensive ranking method, system and device, which can accurately determine the impact of various factor types on documents by extracting multiple factor types from query document pairs and prioritizing them. By assigning a quantitative method of hierarchical weights to factor types, the contribution of each factor to the comprehensive ranking of documents can be more accurately measured, avoiding the need to judge the weights of each factor in different scenarios and different user groups. Retrieval normalization of the factor sequence composed of factor types can improve the accuracy and consistency of retrieval. Retrieval errors caused by inconsistent or non-standard data are reduced. Documents are comprehensively ranked according to the hierarchical weight sequence and the standardized factor sequence. The efficiency and accuracy of information retrieval can be improved, thereby significantly improving the fit between search results and user needs. It enables users to find the relevant documents they need more quickly, greatly improving the user search experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0040] Figure 1 It is a flow chart of the multi-factor comprehensive ranking method provided in the embodiment of the present application.

[0041] Figure 2 It is a structural diagram of a computer system of a server for implementing the method, system, and device embodiments of the present application. DETAILED DESCRIPTION

[0042] The present application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the relevant invention and are not intended to limit the invention. It should also be noted that, for ease of description, only portions relevant to the relevant invention are shown in the accompanying drawings.

[0043] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0044] The embodiments of the present application provide a multi-factor comprehensive ranking method, system, and device.

[0045] In order to more clearly illustrate a multi-factor comprehensive ranking method of the present application, the following Figure 1 Each step in the embodiments of the present application is described in detail.

[0046] The multi-factor comprehensive ranking method of the first embodiment of the present application includes steps S101 to S105, each of which is described in detail as follows:

[0047] S101: Determine a plurality of documents returned based on a document query.

[0048] In this step, based on the document query input by the user, multiple documents related to the document query are retrieved and determined from the document library or data source as an object set for subsequent sorting processing.

[0049] For example, based on the k documents returned by document query q, document query q and one of the k documents form a query-document pair. Document query q has k query-document pairs. The document format for document query q and its corresponding k query-document pairs is {Query_i, Document_1, Document_2,..., Document_k}. Query_i is document query q, and Document_1, Document_2,..., Document_k are the first, second, and kth documents, respectively.

[0050] The following are examples:

[0051] The document query q is: "The current status of academic retrieval development", and three documents are recalled: "The current status of academic retrieval development is one", "The current status of academic retrieval development is two", and "The current status of academic retrieval development is three".

[0052] Based on the above three recall documents, three query document pairs can be constructed, namely:

[0053] {"the current status of academic retrieval development","the current status of academic retrieval development is one"};

[0054] {"the current status of academic retrieval development","the current status of academic retrieval development is two"};

[0055] {“The current status of academic retrieval development”, “The current status of academic retrieval development is three”}.

[0056] Multiple query-document pairs constitute the query-document pair set, which provides the data foundation for the entire multi-factor comprehensive ranking process and clarifies the scope of documents to be ranked. Only with these documents identified can subsequent factor analysis and ranking based on the relationship between the document and the query be performed. Obtaining a set of documents relevant to the query lays the foundation for subsequent factor extraction and ranking, providing a clear target for the ranking operation.

[0057] S102: In a query-document pair set corresponding to a document query, extract multiple factor types from the query-document pair set and prioritize the multiple factor types according to a preset sorting method to obtain a factor type sequence corresponding to the query-document pair set; wherein the query-document pair includes a document query and a document, and a corresponding relationship between the document query and the document.

[0058] In this step, the query-document pair set is composed of multiple query-document pairs. For each query-document pair set, multiple factor types relevant to the query-document pairs are identified and extracted. These extracted factor types are then prioritized according to a preset sorting method to form a factor type sequence unique to the query-document pair set. Within the query-document pair set, each query-document pair shares the multiple extracted factor type sequences.

[0059] For example, the number of factor types extracted from the query document pair set is recorded as m, and the factor types are sorted to obtain a factor type sequence sorted in descending order of priority: T1, T2, ..., Tm.

[0060] In the field of academic information retrieval, users hope to obtain documents with high relevance, high authority and high timeliness. The factor types in this embodiment include at least: relevance type, authority type and timeliness type. The importance ranking of factor types depends on the application scenario. In the field of academic information retrieval, the preset sorting method is: sorting in descending order of priority; specifically: relevance type, authority type and timeliness type. That is, each query document pair in the above-mentioned query document pair set shares the "relevance type, authority type, timeliness type" factor type sequence. In addition, the preset sorting method can be modified according to the adjustment instructions input by the user, and the multi-factor comprehensive sorting of multiple documents can be performed according to the modified sorting method, so that the sorting results of multiple documents are more in line with the user's expectations.

[0061] In this step, by extracting and ranking factor types, the complex relationship between the document and query is converted into an ordered set of factors, facilitating subsequent differentiation and processing of the importance of different factors. This ordered sequence of factor types clearly reflects the relative importance of each factor in the query-document pair. A prioritized sequence of factor types is generated for each query-document pair, allowing the relationship between the document and query to be quantified and analyzed in the form of factors, providing more structured data for subsequent weighting and ranking operations.

[0062] It should be noted that the document query q and one of the k documents form a query-document pair, and the document query q has k query-document pairs. The k query-document pairs constitute the query-document pair set, and the query-document pair set can be extracted from m factor types. For the k query-document pairs of the document query q, the factor types belonging to the document query q form the factor matrix F:

[0063]

[0064] For example: the above document query q is "the current status of academic retrieval development", and the three recalled documents are "the current status of academic retrieval development is one", "the current status of academic retrieval development is two", and "the current status of academic retrieval development is three". The extracted factor types are relevance type, authority type, and timeliness type respectively.

[0065] If the relevance score of the first query document pair is 0.1, the authority score of the first query document pair is 0.11, and the timeliness score of the first query document pair is 0.111; the relevance score of the second query document pair is 0.2, the authority score of the second query document pair is 0.22, and the timeliness score of the second query document pair is 0.222; the relevance score of the third query document pair is 0.3, the authority score of the third query document pair is 0.33, and the timeliness score of the third query document pair is 0.333; then the factor matrix F formed by the document query q is:

[0066]

[0067] Among them, the first row of the factor matrix F (F 11 ,F 12 ,F 13 ,...,F 1m ) is a factor type sequence consisting of m factor types extracted from the document query q, in which the factor types are sorted in descending order of priority: T1, T2, ..., Tm.

[0068] For example, in the example of the document query q being "Current Development Status of Academic Retrieval", the first row of the factor matrix F is the factor sequence of the first query document pair, 0.1 0.11 0.111. From left to right, the relevance score of the first query document pair is 0.1, the authority score of the first query document pair is 0.11, and the timeliness score of the first query document pair is 0.111; the types of each factor in its factor sequence are relevance type, authority type, and timeliness type from left to right; the second and third rows can be obtained by analogy.

[0069] The first column of the factor matrix F (F 11 ,F 21 ,F 31 ,...,F k1 ) is a sequence consisting of the first factor type of the factor type sequence extracted from the k query document pairs of the document query q, whose factor type is T1. The element in the i-th row and j-th column of the factor matrix F is denoted as Fij (i=1,2,...,k; j=1,2,...,m), that is, the factor type of the j-th factor of the i-th query document pair of the document query q is Tj.

[0070] For example, in the above example where the document query q is "the current status of academic retrieval development", 0.1 The first column of F is 0.2, and from top to bottom the relevance scores of the first query document pair are 0.1, 0.3 The relevance score of the second query document pair is 0.2, and the relevance score of the third query document pair is 0.3; each factor type in its factor sequence is relevance type, relevance type, and relevance type from top to bottom; the second and third columns can be obtained by analogy.

[0071] S103: assigning a hierarchical weight to each factor type in the factor type sequence to obtain a hierarchical weight sequence corresponding to the factor type sequence.

[0072] In this step, each factor type in each factor type sequence is assigned a different hierarchical weight based on its importance, further quantifying the priority of the factor type and forming a corresponding hierarchical weight sequence. This resulting hierarchical weight sequence for each query document pair not only establishes a priority order for the factor types but also quantifies their weight values, providing a more detailed and accurate basis for comprehensive ranking.

[0073] As you can see, assigning weights to each factor type allows for a more accurate measurement of its contribution to the document-query relationship. By setting these weights, the influence of different factors can be more accurately reflected in the subsequent comprehensive ranking, making the ranking results more reasonable and scientific.

[0074] In this step, each factor type in the above-mentioned factor type sequence sorted in descending order of priority: T1, T2, ..., Tm is assigned a hierarchical weight, that is, the hierarchical weights are assigned in the following order: HW1, HW2, ..., HW m The weight coefficients assigned to the grading weights must meet the following proportional requirements:

[0075]

[0076] Among them, HW1, HW2, HW3, HW4, HW m-1 and HW m are the weight coefficients corresponding to the 1st, 2nd, 3rd, 4th, m-1st and mth factor types in the factor type sequence respectively (H=BASE^n, BASE=10,11,12,...; n=1,2,3,4,...), and H is the grading coefficient.

[0077] For example, in the example of the document query q being "the current status of development of academic retrieval", the weight coefficients assigned to the factor type sequence "relevance type, authority type, timeliness type" are "100, 10, 1" respectively. The weight coefficients 100, 10, 1 constitute a weight sequence, and the base BASE of the factor type weight sequence is 10.

[0078] S104: Among the multiple factor type sequences, multiple factor sequences are formed based on the same factor type, and each factor sequence is searched and normalized to obtain a normalized factor sequence corresponding to the factor sequence.

[0079] In this step, identical factor types are extracted from multiple factor type sequences to form a single factor sequence. Since multiple identical factor types exist, multiple factor sequences can be extracted. Each factor sequence is then subjected to retrieval normalization, converting it into a standardized factor sequence that meets certain standards or specifications. Retrieval normalization further unifies the representation or processing of these key factors, enabling them to participate in subsequent comprehensive ranking calculations in a more consistent and reliable manner, improving the accuracy and stability of the ranking.

[0080] Specifically, performing retrieval normalization on the factor sequence includes:

[0081] Generate a full range of the factor sequence according to the element value of each factor type in the factor sequence, and perform retrieval normalization on the factor sequence based on the full range.

[0082] It's important to note that the range (R) is a statistic that describes the degree of dispersion in data. It's the difference between the maximum and minimum values in a set of data. The range simply and intuitively reflects the range of fluctuation in the data. A larger range indicates greater dispersion and a more dispersed data distribution; a smaller range indicates more concentrated data and a less dispersed distribution.

[0083] Perform retrieval normalization on the factor sequence, that is, perform retrieval normalization on each column of the factor matrix F. Take the first column of the factor matrix F (F 11 ,F 21 ,F 31 ,...,F k1 ) is recorded as sequence FC1, and the range of sequence FC1 is determined as R = max(FC1) - min(FC1), where max and min are the maximum and minimum values in sequence FC1 respectively. If the range R is 0, no subsequent processing is performed.

[0084] Furthermore, performing search normalization on the factor sequence based on the full range includes:

[0085] Generate a normalization coefficient based on the full range; determine the product of the element value of each factor type in the factor sequence and the normalization coefficient to obtain a normalized sequence; and round down each element value in the normalized sequence to an integer to obtain a normalized factor sequence corresponding to the factor sequence.

[0086] Specifically, if the range R is not 0, a normalization coefficient is generated based on the range. In this embodiment, the ratio of the base BASE of the classification coefficient H to the range R is used as the normalization coefficient.

[0087] Each element value in the canonical sequence is rounded down to an integer, that is, rounded down using the following formula:

[0088] FLOOR(0.1)=FLOOR(0.5)=FLOOR(0.9)=0;

[0089] FLOOR(1.0)=FLOOR(1.5)=FLOOR(1.9)=1.

[0090] It should be noted that FLOOR is a function. The function of the FLOOR function is to round down, that is, to return the largest integer less than or equal to the given value.

[0091] After performing search normalization on the first column of the factor matrix F, the remaining columns are also search normalized to obtain the search normalized factor matrix F * ,as follows:

[0092]

[0093] It should be noted that the factor sequence composed of the factor types in the last column has the smallest weight, so the FLOOR operation is not performed. That is, the value of each element in the factor sequence except the factor sequence with the lowest factor type priority is rounded down to the integer.

[0094] For example, in the example above where the document query q is “the current status of academic retrieval development”, performing retrieval normalization on each column of the factor matrix F yields:

[0095] Multiply the first column by the normalization factor: 10 / (0.3-0.1)=10 / 0.2;

[0096] Multiply the second column by the normalization factor: 10 / (0.33-0.11)=10 / 0.22;

[0097] Multiply the third column by the normalization factor: 10 / (0.333-0.111)=10 / 0.222;

[0098] Determine the product of the element value of each factor type in the factor sequence and the normalized coefficient. The factor matrix F after multiplication is:

[0099]

[0100] Perform FLOOR operation on all columns except the last column which has the smallest weight of factor type to obtain the normalized factor matrix F. * ,as follows:

[0101]

[0102] This step generates a canonical factor sequence, clarifies that the canonical factor sequence is a factor that plays a key role in the comprehensive ranking, and normalizes it, providing a more reliable key factor basis for the final ranking.

[0103] S105: Comprehensively sorting the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence.

[0104] In this step, the multiple documents returned based on the document query, determined in step S101, are comprehensively considered and ranked, combining the hierarchical weight sequence and the normalization factor sequence for each query-document pair to determine their final order. This resulting ranking of the multiple documents, based on a comprehensive consideration of multiple factors, more appropriately meets the user's document query needs and provides a more valuable and relevant document display order.

[0105] Specifically, performing a comprehensive sorting of the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence includes:

[0106] Based on the hierarchical weight sequence HW1, HW2, ..., HWm and the normalization factor sequence after retrieval normalization, a ranking score S is calculated for the k query-document pairs of document query q. The k ranking scores S constitute a scoring sequence. Specifically, the scoring sequence is obtained by matrix multiplying the hierarchical weight sequence with the normalization factor sequence.

[0107] Furthermore, multiple documents are sorted in descending order based on the score sequence, that is, the correspondence between the score sequence and the query document pair, and the elements of the score S are sorted in descending order to obtain an ordered user result list.

[0108] In addition, we can also use the factor matrix F * The scoring matrix S corresponding to the document query q is calculated using the hierarchical weight sequence. The calculation formula of the scoring matrix S is as follows:

[0109]

[0110] Furthermore, multiple documents are sorted in descending order based on the scoring matrix, that is, the correspondence between the scoring matrix and the query document pairs, and the elements of the score S in any column of the scoring matrix are sorted in descending order to obtain an ordered user result list.

[0111] For example, in the above example where the document query q is “the current status of academic retrieval development”, the factor matrix F * The score for the first query-document pair calculated using the hierarchical weight sequence is: 5x100 + 5x10 + 5x1 = 555. The score for the second query-document pair is: 10x100 + 10x10 + 10x1 = 1100. The score for the third query-document pair is: 15x100 + 15x10 + 15x1 = 1665. The result list returned to the user in descending order of score is: {"The current status of academic search development is three," "The current status of academic search development is two," "The current status of academic search development is one"}.

[0112] This step comprehensively considers the weights of different factor types and the standardized representation of key factors, and incorporates various relationships and important factors between documents and queries into the ranking calculation, so that the ranking results can more comprehensively and accurately reflect the relevance and importance of documents and queries.

[0113] As can be seen from the above description, a multi-factor comprehensive ranking method provided by an embodiment of the present application can accurately determine the impact of various factor types on documents by extracting multiple factor types from query document pairs and prioritizing them. By assigning a quantitative method of hierarchical weights to factor types, the contribution of each factor to the comprehensive ranking of documents can be more accurately measured, avoiding the need to judge the weights of each factor in different scenarios and different user groups. Retrieval normalization of the factor sequence composed of factor types can improve the accuracy and consistency of retrieval. Retrieval errors caused by inconsistent or non-standard data are reduced. Documents are comprehensively ranked according to the hierarchical weight sequence and the standardized factor sequence. The efficiency and accuracy of information retrieval can be improved, thereby significantly improving the fit between search results and user needs. It enables users to find the relevant documents they need more quickly, greatly improving the user search experience.

[0114] Although the various steps in the above embodiment are described in the above-mentioned order, those skilled in the art will understand that in order to achieve the effect of this embodiment, different steps do not have to be executed in such an order. They can be executed simultaneously (in parallel) or in a reverse order. These simple changes are within the scope of protection of the present invention.

[0115] A second embodiment of the present invention provides a multi-factor comprehensive ranking system, comprising:

[0116] An acquisition module for determining multiple documents returned based on a document query;

[0117] a priority module configured to extract, from a query-document pair set corresponding to a document query, a plurality of factor types and prioritize the plurality of factor types according to a preset sorting method to obtain a factor type sequence corresponding to the query-document pair set; wherein the query-document pair comprises a document query and a document, and a corresponding relationship between the document query and the document;

[0118] A weight module, configured to assign a hierarchical weight to each factor type in the factor type sequence to obtain a hierarchical weight sequence corresponding to the factor type sequence;

[0119] a processing module, configured to construct multiple factor sequences based on the same factor type from the multiple factor type sequences, and perform search and normalization on each of the factor sequences to obtain a normalized factor sequence corresponding to the factor sequence;

[0120] The sorting module is used to comprehensively sort the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence.

[0121] In an optional embodiment, the processing module includes:

[0122] a calculation unit, configured to generate a full range of a factor sequence according to an element value of each factor type in the factor sequence;

[0123] A normalization unit is used to perform retrieval normalization on the factor sequence based on the full range.

[0124] In an optional embodiment, the normalization unit includes:

[0125] a coefficient subunit, configured to generate a normalized coefficient based on the full range;

[0126] a product subunit, configured to determine the product of the element value of each factor type in the factor sequence and the normalization coefficient to obtain a normalized sequence;

[0127] The normalization subunit is used to round down each element value in the normalized sequence to an integer to obtain a normalized factor sequence corresponding to the factor sequence.

[0128] In an optional embodiment, the sorting module includes:

[0129] A scoring unit, configured to generate a scoring sequence according to the hierarchical weight sequence and the normalization factor sequence;

[0130] A sorting unit is used to sort the multiple documents in descending order based on the scoring sequence.

[0131] In an optional embodiment, the scoring unit includes:

[0132] The scoring subunit is used to perform matrix multiplication on the hierarchical weight sequence and the normalization factor sequence to obtain a scoring sequence.

[0133] In an optional embodiment, the weight coefficients assigned to the hierarchical weights meet the following proportional requirements:

[0134]

[0135] Among them, HW1, HW2, HW3, HW4, HWm-1 and HWm are the weight coefficients corresponding to the factor types of the 1st, 2nd, 3rd, 4th, m-1 and mth items in the factor type sequence respectively, and H is the grading coefficient.

[0136] In an optional embodiment, the factor types include at least: relevance type, authority type and timeliness type.

[0137] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process and related instructions of the system described above can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0138] It should be noted that the multi-factor comprehensive ranking system provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be decomposed or combined. For example, the modules in the above embodiment can be combined into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the modules or steps and are not regarded as improper limitations of the present invention.

[0139] An electronic device according to a third embodiment of the present invention includes:

[0140] at least one processor;

[0141] and a memory communicatively coupled to at least one of said processors;

[0142] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the above-mentioned multi-factor comprehensive ranking method.

[0143] A fourth embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the above-mentioned multi-factor comprehensive ranking method.

[0144] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes and related instructions of the storage device and processing device described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0145] Those skilled in the art should be able to appreciate that, in conjunction with the modules and method steps of each example described in the embodiments disclosed herein, it is possible to implement them with electronic hardware, computer software, or a combination of the two, and the programs corresponding to the software modules and method steps can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. In order to clearly illustrate the interchangeability of electronic hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0146] Reference below Figure 2 , which shows a structural diagram of a computer system of a server for implementing the method, system, and device embodiments of the present application. Figure 2 The server shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0147] like Figure 2 As shown, the computer system includes a central processing unit (CPU) 201, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 202 or the program loaded from the storage part 208 to the random access memory (RAM) 203. Various programs and data required for system operation are also stored in the RAM 203. The CPU 201, ROM 202, and RAM 203 are connected to each other via a bus 204. An input / output (I / O) interface 205 is also connected to the bus 204.

[0148] The following components are connected to the I / O interface 205: an input section 206 including a keyboard, a mouse, and the like; an output section 207 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 208 including a hard disk and the like; and a communication section 209 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 209 performs communication processing via a network such as the Internet. A drive 210 is also connected to the I / O interface 205 as needed. A removable medium 211, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 210 as needed, so that a computer program read therefrom can be installed into the storage section 208 as needed.

[0149] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 209, and / or installed from the removable medium 211. When the computer program is executed by the central processing unit (CPU) 201, the above-mentioned functions defined in the method of the present application are executed. It should be noted that the computer-readable medium mentioned above in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code embodied on a computer-readable medium may be transmitted using any suitable medium, including, but not limited to, wireless, wire, optical cable, RF, etc., or any suitable combination thereof.

[0150] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0151] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0152] The terms "first", "second", etc. are used to distinguish similar objects, rather than to describe or indicate a particular order or sequence.

[0153] The term "comprise" or any other similar term is intended to cover non-exclusive inclusion such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0154] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A multi-factor comprehensive ranking method, characterized in that: include: Determine multiple documents to be returned based on a document query; In a query-document pair set corresponding to a document query, extracting multiple factor types from the query-document pair set and prioritizing the multiple factor types according to a preset sorting method to obtain a factor type sequence corresponding to the query-document pair set; wherein the query-document pair includes a document query and a document, and a corresponding relationship between the document query and the document; Assigning a hierarchical weight to each factor type in the factor type sequence to obtain a hierarchical weight sequence corresponding to the factor type sequence; In the plurality of factor type sequences, a plurality of factor sequences are formed based on the same factor type, and each of the factor sequences is searched and normalized to obtain a normalized factor sequence corresponding to the factor sequence; The returned multiple documents are comprehensively sorted according to the hierarchical weight sequence and the normalization factor sequence.

2. The multi-factor comprehensive ranking method according to claim 1, characterized in that: The factor sequence is searched and normalized, including: Generate the full range of the factor sequence according to the element value of each factor type in the factor sequence; The factor sequence is searched and normalized based on the full range.

3. The multi-factor comprehensive ranking method according to claim 2, characterized in that: The step of retrieving and normalizing the factor sequence based on the full range includes: generating a normalized coefficient based on the full range; Determine the product of the element value of each factor type in the factor sequence and the normalization coefficient to obtain a normalized sequence; Each element value in the canonical sequence is rounded down to an integer to obtain a canonical factor sequence corresponding to the factor sequence.

4. The multi-factor comprehensive ranking method according to claim 1, characterized in that: The comprehensively sorting the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence includes: generating a scoring sequence according to the hierarchical weight sequence and the normalization factor sequence; The plurality of documents are sorted in descending order based on the score sequence.

5. The multi-factor comprehensive ranking method according to claim 4, characterized in that: Generating a scoring sequence according to the hierarchical weight sequence and the normalization factor sequence includes: The hierarchical weight sequence and the normalization factor sequence are matrix multiplied to obtain a scoring sequence.

6. The multi-factor comprehensive ranking method according to claim 1, characterized in that: The weight coefficients assigned to the hierarchical weights meet the following proportional requirements: Among them, HW1, HW2, HW3, HW4, HW m-1 and HW m are the weight coefficients corresponding to the 1st, 2nd, 3rd, 4th, m-1 and mth factor types in the factor type sequence, and H is the grading coefficient.

7. The multi-factor comprehensive ranking method according to claim 1, characterized in that: The factor types include at least: relevance type, authority type and timeliness type.

8. A multi-factor comprehensive ranking system, characterized in that: include: An acquisition module for determining multiple documents returned based on a document query; a priority module configured to extract, from a query-document pair set corresponding to a document query, a plurality of factor types and prioritize the plurality of factor types according to a preset sorting method to obtain a factor type sequence corresponding to the query-document pair set; wherein the query-document pair comprises a document query and a document, and a corresponding relationship between the document query and the document; A weight module, configured to assign a hierarchical weight to each factor type in the factor type sequence to obtain a hierarchical weight sequence corresponding to the factor type sequence; a processing module, configured to construct multiple factor sequences based on the same factor type from the multiple factor type sequences, and perform search and normalization on each of the factor sequences to obtain a normalized factor sequence corresponding to the factor sequence; The sorting module is used to comprehensively sort the returned multiple documents according to the hierarchical weight sequence and the normalization factor sequence.

9. An electronic device, characterized in that: include: at least one processor; and a memory communicatively coupled to at least one of said processors; The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the multi-factor comprehensive ranking method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the multi-factor comprehensive ranking method according to any one of claims 1 to 7.

Images