Optimization method for index retrieval of digital library

Abstract

The invention relates to an optimization method of retrieval, in particular to an optimization method of index retrieval of a digital library, which adopts the following technical scheme: S1, a dynamic hierarchical index construction technology: firstly, performing multi-dimensional content analysis on digital resources in the library, including themes, keywords, authors and publishing time information; classifying the literatures according to the information, and constructing a content hierarchical structure; s2, an index optimization strategy based on a use mode: performing data analysis on a retrieval behavior of a user, monitoring an access frequency, a query type and an access path of the user in different types of literatures, and identifying a typical user use mode; s3, performing semantic association retrieval optimization; and S4, index updating and resource allocation optimization: when new literature resources are added into a library or existing resources are modified, updating the index in real time, and adjusting an index structure according to content classification and user behaviors.

Description

Technical Field

[0001] The invention relates to a retrieval optimization method, in particular to an index retrieval optimization method for a digital library. Background Art

[0002] Among the current digital library index retrieval optimization methods, the existing technology represented by the "Digital Library Content Analysis System and Method" with Chinese invention patent number 2017103045690 does have advantages in many aspects, especially in content analysis and user input processing. However, with the increasing complexity of user needs and the rapid growth of information volume, this method still has deficiencies and drawbacks in many aspects, especially in key areas such as large-scale data processing, personalized retrieval and semantic analysis. These problems directly affect the index retrieval optimization effect of digital libraries.

[0003] First, the existing index retrieval system lacks effective use of user behavior feedback, especially in the use of data such as click-through rate and dwell time. The system in Patent No. 2017103045690 can extract user input information and perform preliminary analysis through signal reception and judgment steps, but its core still relies on keyword matching and classification models. Although different contents are grouped in the classification step and attempts are made to model by content characteristics, the optimization effect of user behavior feedback on the retrieval system is not deeply considered. For example, user click-through rate and dwell time are often important indicators for measuring the value and relevance of documents, but the existing system does not use these behavioral data to dynamically adjust the index priority and model parameters. This makes it difficult for the system to be personalized according to the actual needs of users, the accuracy and relevance of the retrieval results are insufficient, and the user experience is limited. Secondly, the existing technology has serious limitations in semantic understanding and context analysis. The method in Patent No. 2017103045690 mainly relies on the first matching device and the second matching device to perform preliminary matching of content, but its matching method still basically relies on direct matching of keywords. Although the system can analyze and group the content input by the user through the signal receiving step and the classification step, it cannot deeply understand the semantic intent behind the query. Especially when the user inputs a query with complex semantics, the existing system cannot perform contextual analysis, resulting in the matching results being too mechanical and unable to capture the user's real needs. The lack of semantic matching not only reduces the accuracy of the retrieval results, but also makes the system perform poorly when facing long queries or semantically ambiguous queries, and is prone to the problem of irrelevant documents or results deviation.

[0004] Third, the existing system lacks support for personalized user needs and the ability to adjust dynamically in real time. Users of digital libraries often have their own unique needs and interests. Although the classification devices in existing patents can group and analyze based on content characteristics, they fail to fully consider the personalized needs and long-term behavioral habits of users. Over time, users' interests and behavior patterns will change, but the index optimization process of the system is static, and the index results will not be adjusted in real time due to long-term behavioral feedback from users. This design limits the system's adaptive ability and cannot dynamically optimize the index structure according to the user's behavior pattern, resulting in the system being unable to achieve accurate and fast response when facing a large number of users searching at the same time.

[0005] In addition, although the signal receiving and judgment steps in the prior art have improved the speed of information extraction to a certain extent, they are not ideal in large-scale data processing and real-time index updating. With the continuous growth of the amount of documents in digital libraries, it is difficult for existing systems to effectively process massive data in a short period of time, especially in real-time indexing and rapid response. The processing speed of the system is difficult to meet the user's efficient retrieval needs. For example, when the second matching device faces a large amount of user input, it is easy for the system response speed to decrease due to excessive processing time. This not only increases the user's waiting time, but may also miss some high-value instant information. This processing bottleneck is particularly prominent when dealing with modern large-scale data retrieval. Summary of the invention

[0006] The purpose of the present invention is to provide an optimization method for index retrieval of a digital library, thereby solving some of the drawbacks and shortcomings pointed out in the background technology.

[0007] The present invention solves the above-mentioned technical problems by adopting the following technical solutions:

[0008] S1. Dynamic hierarchical index construction technology:

[0009] S1.1. First, conduct a multi-dimensional content analysis of the digital resources in the library, including subject, keyword, author, and publication time information; classify the documents based on the information and construct a content hierarchy;

[0010] S1.2. Based on the classification, dynamic indexes are constructed for different document levels including topics, content details, and semantic associations, so that commonly used high-frequency content can be quickly retrieved, while detailed content is retained in the deep index;

[0011] S1.3. Regularly adjust the index structure based on the user's usage frequency and search habits, so that commonly used information is close to the surface and the search speed is optimized;

[0012] S2. Index optimization strategy based on usage pattern:

[0013] S2.1. By analyzing the user's search behavior, monitoring the user's access frequency, query type and access path in different categories of documents, and identifying typical user usage patterns;

[0014] S2.2. Dynamically adjust the priorities of different index items based on user behavior monitoring results, and promote frequently accessed documents or related topics to the top of the search results;

[0015] S2.3. Cache the high-frequency index parts locally according to the usage pattern, return the results quickly, and control and reduce the resource usage;

[0016] S3, semantic association retrieval optimization:

[0017] S3.1. Introduce natural language processing technology to perform semantic analysis on user query statements, identify keywords and understand the semantic intent behind them, and expand the index scope;

[0018] S3.2, based on semantic analysis, intelligently match the user's query with the semantics of the document content, optimize the index, and enable documents with similar semantics but different keywords to be retrieved;

[0019] S3.3. Optimize the semantic matching model and index structure based on the click-through rate of search results and user dwell time feedback;

[0020] S4. Index update and resource allocation optimization:

[0021] S4.1. Update the index in real time when new literature resources are added to the library or existing resources are modified, and adjust the index structure according to content classification and user behavior;

[0022] S4.2, using distributed database technology to store indexes on multiple nodes and support large-scale parallel retrieval;

[0023] S4.3. During the retrieval process, resources are dynamically allocated and each retrieval node is scheduled according to the current query volume and load conditions.

[0024] Furthermore, the method for constructing the dynamic index includes:

[0025] S1. First, we use content analysis and classification algorithms to conduct multi-dimensional analysis of digital resources in the library, including subject, content details, and semantic association information, and establish a preliminary structure of a dynamic index based on document hierarchy. For each document, its index is calculated hierarchically using the formula:

[0026]

[0027] Among them, S(x) represents the dynamic index score of the document, which is the core variable used to measure the importance of the document in the index optimization process; f1(x) is the subject relevance function of the document, which reflects the degree of match between the document content and a specific subject; f2(x) is the content detail function of the document, which represents the richness of the description of the document at the detail level; α is the adjustment coefficient of the content details, which is used to adjust the weight of the details in the overall index; β(t) is the time dynamic adjustment coefficient related to user interaction, which dynamically adjusts the index structure over time; T is the time period for index weight adjustment, which is used to represent the time window for calculating the index in the optimization process;

[0028] S2. Secondly, by monitoring user behavior, including query frequency, access time, and user path, the frequency perception model is used to dynamically adjust the index position of the document. The weight adjustment of the index is calculated by the formula:

[0029]

[0030] Among them, W(u) is the index weight adjustment result caused by user behavior, which indicates the index priority after the user visits a specific document; p i is the access frequency of the ith document, that is, the number of times the document is queried by users; λ is the adjustment parameter that controls the frequency sensitivity of user behavior, which determines the intensity of the impact of access frequency on the document index weight; t represents the current query time, which adjusts the perception of document access frequency over time;

[0031] S3. Finally, through the semantic association analysis technology, deep semantic matching of documents is carried out. The semantic vector model based on deep learning is used to dynamically optimize the semantic level matching of different document contents. The semantic level is determined by the formula:

[0032]

[0033] Among them, L(s) represents the semantic level score of the document, which is the result of dividing the document into levels through semantic analysis; g1(s) is the basic semantic matching function, which represents the basic semantic association of the document; g2(s) is the semantic extension function, which is responsible for matching the document in the semantic space; γ is the semantic extension coefficient, which controls the impact of semantic extension on the matching results; n is the detail complexity of the document, which is used to measure the complexity of the detailed description of the document at the semantic level.

[0034] Furthermore, the method for identifying a typical user usage pattern includes:

[0035] S1. First, log tracking and Cookie technology are used to record the user's search behavior data in real time, including the document category, query type, access path and stay time of each visit, and the user behavior data is modeled and optimized in multiple dimensions using the formula:

[0036]

[0037] Among them, B(x) represents the comprehensive score of user behavior; f1(x) is the access frequency function, which represents the frequency of user access to a certain document; f2(x) is the residence time function, which reflects the length of time the user stays in the document; ω1 and ω2 are the weight factors of access frequency and residence time respectively; ψ(t) is the behavior dynamic adjustment function, which affects the index priority over time; T is the behavior time window for real-time monitoring;

[0038] S2. Then, based on the collected behavioral data, cluster analysis algorithms such as K-Means or DBSCAN are used to classify users, identify typical user usage patterns, including users who frequently query specific categories of documents or often follow similar paths, and mathematically model these patterns;

[0039] Then, based on the user's historical behavior data, a deep learning model based on the recurrent neural network (RNN) is constructed to predict the user's future retrieval needs through the following formula:

[0040]

[0041] Among them, P(u) is the prediction of the user's future behavior; u i represents the ith retrieval behavior, f3(u i ) is the user behavior sequence feature function, which describes the user's past retrieval behavior pattern; λ is the adjustment parameter that controls the prediction sensitivity, t i Indicates the time when the behavior occurs.

[0042] Furthermore, based on the prediction results of P(u), the index of relevant documents is loaded in advance, and the retrieval efficiency is optimized using an adaptive preloading mechanism; then, according to the actual feedback of user behavior, the display order of the index is adjusted, and the dynamic sorting of the index is optimized using the formula:

[0043]

[0044] Among them, S(y) represents the priority ranking score of the document in the index; r1(y i ) and r2(y i ) are the user's query frequency and path click rate function in the document, γ1 and γ2 are adjustment coefficients, which control the influence of the two on the priority.

[0045] Furthermore, the framework method used to identify keywords and understand the semantic intent behind them is:

[0046] S1. First, use the deep learning natural language processing model to perform deep semantic analysis on the user query, and use the following formula to calculate the context semantic vector of the user query:

[0047]

[0048] Where V(q) represents the contextual semantic vector of the query statement, which is used to quantify the semantic content of the user query; f1(w i ) is the keyword semantic weight function, which represents the i-th keyword w in the query i The weight in the semantic space reflects the relevance of the keyword to the query topic; f2(c i ) is the context-dependent function, indicating that keyword w i With its context c i The semantic relationship between them; α and β are adjustment coefficients, which are used to balance the weights of keywords and context; γ(t) is a time dynamic adjustment function, which represents the semantic characteristics of the query changing over time; T is the time window of semantic parsing, which represents the duration of semantic parsing;

[0049] S2. Then, we use semantic intent recognition technology to classify user queries into different types through the intent classification model, including information queries, navigation queries, and transaction queries, and use the formula to construct a multi-level index for the intent:

[0050]

[0051] Among them, I(s) represents the hierarchical score of query intent, which is an indicator used to quantify user query intent; g i (x) is a semantic level function, which indicates the score of document x in different intent levels and reflects the semantic matching degree between document and user query; λ is a parameter that controls the time sensitivity of the model and determines the speed at which query intent changes over time; t i is the timestamp of the query;

[0052] S3. Finally, through the dynamic semantic expansion mechanism, the knowledge graph or deep semantic model is used to expand keywords according to the semantic intent of the user's current query statement. The calculation formula is:

[0053]

[0054] Among them, R(q) is the recommended search result set after expansion, which means the final search result after semantic expansion; h1(y i ) is the direct keyword matching result function, which represents the document y directly related to the query q i score; h2(y i ) is a semantic expansion result function, which represents documents y that are semantically related to query q but not directly keyword matched.i score; μ is the semantic expansion coefficient, which controls the influence of the expansion result on the final recommendation; n is the total number of retrieved documents.

[0055] Furthermore, the optimization of the semantic matching model and the index structure includes the following steps:

[0056] S1. First, collect the user's click behavior and dwell time in the search results in real time, associate the data with the semantic relevance of the document, and calculate the impact of user behavior feedback on the semantic priority of the document based on the formula:

[0057]

[0058] Where P(u) represents the semantic priority score of document u, C u (t) is the change of the click rate of the document over time t, S u (t) is the user residence time function of the document, α and β are adjustment coefficients that control the relative importance of click-through rate and residence time, λ is the time decay factor, and T is the collection time period of feedback data;

[0059] S2. Then, the behavior feedback is input into the semantic matching model to dynamically adjust the position of the document in the semantic space. The deep learning algorithm is used to analyze the long-term behavior and optimize the parameters of the semantic model. Then, based on the user's stay time and click-through rate, the reinforcement learning algorithm is used to dynamically adjust the index structure and update the index priority of the document through the formula:

[0060]

[0061] Among them, I(v) is the priority score of document v in the index, S i (v) is the residence time of the document at the time of the i-th search, C i (v) is the i-th click rate, μ is the coefficient of the control feedback on the priority adjustment, t i is the timestamp of the action;

[0062] S3. Finally, we optimize the semantic extension model based on behavioral feedback, dynamically optimize the extended keyword pool based on user clicks and dwell time, and use the formula to calculate personalized semantic extension recommendations:

[0063]

[0064] Among them, E(q) is the semantic expansion score of query q, f1(w i ) is the keyword w i The expansion weight, f2(c i ) is the context semantic expansion function, k1 and k2 are expansion parameters that control the relative influence of keywords and context, and t is the time when the query occurs.

[0065] The beneficial effects of the present invention are as follows: by dynamically adjusting the index priority of documents and using an adaptive preloading mechanism, the system can preload the index of relevant documents according to the user's query history, behavioral feedback, and semantic matching, thereby significantly improving the retrieval speed and reducing the user's waiting time.

[0066] By combining user click-through rate, dwell time and other behavioral data, the semantic priority of documents is optimized in real time, so that documents that best match user needs are displayed first. Through the semantic expansion mechanism, the system can also find documents that are semantically related but not completely matched, further enriching the search results and ensuring that users can get a more accurate search experience.

[0067] Based on deep learning and reinforcement learning algorithms, the system can dynamically adjust the index structure according to the user's long-term behavior patterns to ensure personalized document recommendations. At the same time, the semantic expansion mechanism optimizes the keyword pool by analyzing the user's clicks and dwell time, and provides personalized semantic expansion recommendations to meet the search needs of different users. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a flow chart of the optimization method for index retrieval of a digital library of the present invention.

[0069] Figure 2 A flow chart of the method used to construct a dynamic index for the present invention.

[0070] Figure 3 A flow chart of a method for identifying typical user usage patterns is provided for the present invention.

[0071] Figure 4 A flow chart of the framework method used by the present invention to identify keywords and understand the semantic intent behind them.

[0072] Figure 5 This is a flow chart of the method for optimizing the semantic matching model and index structure of the present invention. DETAILED DESCRIPTION

[0073] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0074] Combined with Figure 1 , improve retrieval efficiency and user experience through dynamic hierarchical index construction technology.

[0075] First, in step S1.1, the system conducts a multi-dimensional content analysis of the digital resources in the library, and the dimensions of the analysis include key information such as the subject, keywords, author, and publication time of the document. This information is used to classify the documents, thereby building a content hierarchy so that subsequent retrieval operations can organize and manage document data more efficiently.

[0076] Based on the classification, in step S1.2, a dynamic index structure is constructed for different levels of documents, such as topics, content details, and semantic associations. This dynamic index structure is divided into two parts: shallow and deep. Commonly used high-frequency content is placed in the shallow index for quick retrieval, while more detailed content is retained in the deep index, waiting to be called when more complex or specific queries are made. The design of this index structure ensures that commonly used information can be retrieved quickly without affecting the overall efficiency due to the system processing detailed information.

[0077] In order to further optimize the user's search experience, in step S1.3, the system will regularly dynamically adjust the index structure based on the user's usage frequency and search habits. By analyzing the user's usage data, the system can place the documents or common information that the user frequently accesses closer to the surface of the index, so that these high-frequency contents can be retrieved by the user more quickly. At the same time, the system will continue to monitor the user's behavior to ensure that the index structure is always in the best state.

[0078] Improve retrieval efficiency and user experience through index optimization strategies based on usage patterns.

[0079] First, in step S2.1, the system conducts a comprehensive data analysis of the user's search behavior to monitor the user's access frequency, query type, and access path in different categories of documents. This includes recording the user's query history, click behavior, and browsing path data to identify typical user usage patterns. These usage patterns not only reflect the user's preferences, but also reveal the types and topics of documents that the user frequently visits during the search process.

[0080] In step S2.2, the system dynamically adjusts the priority of different index items based on the user behavior monitoring results. Through this process, the system can automatically promote documents or related topics that users frequently visit to the top of the search results, thereby ensuring that these documents can be retrieved first when users perform similar queries. The dynamic adjustment mechanism in this step enables the system to adapt to changes in user needs and continuously optimize the display order of search results, thereby improving the relevance of search results and user satisfaction.

[0081] Next, in step S2.3, the system caches the frequently accessed index parts locally based on the user's usage pattern. This caching mechanism can greatly improve the retrieval speed, because when the user issues a query, the system can quickly return the results directly from the cache without having to retrieve from the global index every time. In addition, the local cache can also effectively control and reduce resource usage, allowing the system to maintain low resource consumption while providing efficient retrieval services.

[0082] Through the semantic association retrieval optimization strategy, the relevance and accuracy of the retrieval results are improved.

[0083] First, in step S3.1, the system introduces natural language processing (NLP) technology to perform deep semantic analysis on the user's query. This analysis not only identifies the keywords entered by the user, but also understands the semantic intent behind the query by analyzing the context structure and context of the sentence. In this way, the system can expand the scope of the index, no longer limited to the surface keywords entered by the user, but combine semantic intent to perform a broader index expansion and capture the user's potential retrieval needs.

[0084] Then, in step S3.2, the system intelligently matches the user's query with the semantics of the document content based on semantic analysis. Through the deep semantic understanding model, the system can identify the parts of the document that are semantically similar to the user's query, even if the keywords of these documents do not completely match the user's query. Through this semantic-level optimization, the system can more accurately retrieve documents that meet the user's actual needs, rather than just those results with the same keywords. Especially when the user uses complex queries, semantic matching can greatly improve the accuracy of the retrieval.

[0085] Finally, in step S3.3, the system continuously optimizes the semantic matching model and index structure based on the click-through rate and user dwell time feedback of the search results. Click-through rate and dwell time are important behavioral data for measuring the relevance of documents. By analyzing these feedbacks in real time, the system can dynamically adjust the weights and parameters of the semantic matching model to give priority to documents with long user dwell time and high click-through rate. At the same time, the system will also optimize the index structure based on these behavioral feedbacks, so that future search results are more in line with the interests and needs of users.

[0086] Through index update and resource allocation optimization strategies, we ensure that the system can efficiently handle the addition and modification of new documents, while improving the performance of large-scale retrieval.

[0087] In step S4.1, when new document resources are added to the library or existing resources are modified, the system will update the index in real time. This process is completed through an automated index update mechanism to ensure that the content of each document is kept up to date in the index. In addition, the system also adjusts the index structure based on content classification and user behavior data, so that commonly used and frequently accessed documents can be prioritized at the top of the index to improve retrieval speed and user experience.

[0088] In step S4.2, the system uses distributed database technology to store index data on multiple nodes. This architecture design can support large-scale parallel search operations. When users initiate a large number of search requests at the same time, the system can make full use of the advantages of distributed databases and schedule multiple nodes to work together to ensure load balancing of search tasks and avoid performance bottlenecks of single nodes.

[0089] Finally, in step S4.3, the system dynamically allocates resources and intelligently schedules each search node based on the current query volume and load. When the system detects that the load on some nodes is too high, it can automatically allocate some query tasks to other nodes with lower loads, thereby optimizing the overall performance of the system. This dynamic scheduling mechanism for resource allocation can not only ensure that the system remains efficient under high concurrency conditions, but also optimize the system's response time, ensuring that users can obtain search results in a short time.

[0090] Embodiment 1:

[0091] Combined with Figure 2 Flowchart,In a large digital library, the document library contains more than five million academic papers and documents, covering multiple disciplines, including physics, chemistry, biology, computer science, etc. In order to improve the user's search experience, the library decided to adopt a dynamic index optimization method to optimize the index structure according to the different query needs of users, so as to quickly retrieve the most relevant documents.

[0092] Suppose a user searches for the topic "Application of artificial intelligence in medicine" in the library. The library's indexing system will analyze and dynamically sort all relevant documents to ensure that documents with high relevance and detailed content are displayed first in the search results.

[0093] First, the system uses content analysis and classification algorithms to conduct multi-dimensional analysis of digital resources in the library. For example, for the literature related to the topic "Application of artificial intelligence in medicine", the system will analyze it from three dimensions:

[0094] 1. Topic relevance: frequency and importance of keywords such as “artificial intelligence”, “medical care”, “diagnostic system”, “machine learning”, etc.

[0095] 2. Content details: whether the literature contains detailed algorithm descriptions, specific cases, experimental data, research methods, etc.

[0096] 3. Semantic relevance: The system will understand the context of the document through natural language processing technology and analyze the potential semantic relationships in the document to determine which documents have content closely related to the topic, although they may not have direct keyword matches.

[0097] The analysis results will construct a preliminary hierarchy of the literature and perform preliminary calculations of the dynamic index based on S(x) in the formula.

[0098] Next, based on the previous analysis, the system will use the following formula to calculate the dynamic index score of the document:

[0099]

[0100] Among them, assuming that when a user searches for "application of artificial intelligence in medicine", the relevance and content analysis results of a certain document x1 are as follows:

[0101] 1. Topic relevance function f1(x1): reflects the degree of match between the document and the topics of "artificial intelligence" and "medical care". Assuming that this document contains a lot of discussion on "artificial intelligence applied to medical diagnosis systems", its topic relevance is 0.85.

[0102] 2. Content detail function f2(x1): The document contains detailed machine learning model descriptions and clinical trial data, with high detail richness. Assume that the detail score is 0.75.

[0103] 3. Adjustment coefficient α: Since libraries give priority to documents with rich details, it is assumed that α is 1.2, which indicates the system's preference for content details.

[0104] 4. Time dynamic adjustment coefficient β(t): Assuming that the user has searched for similar topics multiple times in the past month, the system will dynamically increase the priority of documents related to the user's historical queries. Assume that the initial value of β(t) is 1.1 and gradually increases over time.

[0105] Substitute these data into the formula to calculate the index score of the document. Assuming the time window T is 30 days, then:

[0106]

[0107] The calculations show that:

[0108]

[0109] S(x1)=1.84×30=55.2

[0110] Therefore, the dynamic index score of document x1 is 55.2.

[0111] In order to ensure that the system can quickly respond to user query needs, the system will regularly adjust the index structure according to the user's usage frequency and search habits. Assuming that the user clicked on the literature related to "AI medical applications" in the recent searches and stayed for a long time, these behavioral data will be fed back to the system, and the system will further increase the priority of these documents. Assuming that document x1 is visited by the user many times and has a high click-through rate, the system will place this document in a higher position in the next query, thereby optimizing the user's search experience.

[0112] In this example, the index score of document x1 is 55.2. Based on this score, the system places it in a higher priority position, ensuring that users can quickly find this highly relevant and detailed document when searching for "application of artificial intelligence in medicine".

[0113] It is assumed that user behavior monitoring is used to further optimize the index position of documents.

[0114] Users frequently search for literature related to "Application of Artificial Intelligence in Medicine" in the library. The system will monitor the user's query frequency, access time, and browsing path between literature in real time. Based on these behavioral data, the system will dynamically adjust the indexing priority of the literature through the frequency perception model.

[0115] In the previous search process, the dynamic index score of document x1 was 55.2, and the system prioritized it to users based on this score. Now, the system will further adjust the index weights of document x1 and other related documents based on user behavior feedback to ensure that documents that users visit more frequently can be displayed more quickly in future searches.

[0116] Assume that the user has searched the topic "Application of artificial intelligence in medicine" several times in the past month, and the access behavior on documents x1, x2, and x3 is recorded by the system. Specifically, the user clicked on document x1 10 times, document x2 7 times, and document x3 5 times. In addition, the user stayed on document · for a long time, indicating the high relevance of the document.

[0117] In order to dynamically adjust the priority of documents according to user behavior, the system uses a frequency-aware model to calculate the weight of documents. The formula is:

[0118]

[0119] Assume that λ is a tuning parameter that controls the frequency sensitivity of user behavior, and its value range is usually between 0.1 and 0.5. Assume that the system sets λ = 0.3 to moderately respond to changes in user behavior. At the same time, t represents the current query time, for example, set to 30 days.

[0120] For document x1, its access frequency p1 is 10 times, the access frequency p2 of document x2 is 7 times, and the access frequency p3 of document x3 is 5 times. Substitute these data into the formula for calculation:

[0121]

[0122] It can be seen from the calculation that since the query time t is relatively short, the exponential part e in the formula -λt will quickly approach 0, causing the index weight to be almost equal to the access frequency p i Therefore, the index weight of document x1 is 10, document x2 is 7, and document x3 is 5.

[0123] Based on these weighted results, the system will adjust the index position of the document. In the future, when users perform similar searches, document x1 will be displayed first because it has the highest weight, and documents x2 and x3 will also be displayed in descending order of weight.

[0124] Assuming that another new user also searches for the topic "Application of artificial intelligence in medicine", the system will automatically rank document x1 at the top of the search results. Because the previous user's behavior data has shown that the document is highly relevant and rich in content, the new user can quickly find the most useful document, improving the overall user search experience. As more and more user behaviors are monitored and fed back to the system, the system will continuously adjust the index weight, maintain dynamic updates and optimizations, so that frequently accessed documents are always displayed first.

[0125] In the previous calculation, documents x1, x2, and x3 have already received higher index weights through user behavior feedback. However, in this scenario, users want to query documents with more semantic depth, such as "specific applications of artificial intelligence in medical diagnosis and treatment", which requires the system to not only match keywords, but also understand the semantic intent behind the document content. Therefore, the system will dynamically calculate the semantic level score of the document through semantic association analysis technology, combining the user's query and document content.

[0126] The system uses the following formula to optimize the semantic level matching of documents:

[0127]

[0128] in:

[0129] 1. L(s) is the semantic level score of the document, which measures the degree of semantic matching of the document.

[0130] 2. g1(s) is the basic semantic matching function, which indicates the basic semantic relevance between the document and the user's query topic. For example, the number and importance of keywords such as "artificial intelligence", "medical", and "diagnosis" in the document.

[0131] 3.g2(s) is a semantic expansion function, which is responsible for expanding the matching semantics according to the context and identifying document content with potential relevance.

[0132] 4.γ is the semantic expansion coefficient, ranging from 0.5 to 1.5, which is used to adjust the impact of semantic expansion on the final matching results. For example, when γ is large, semantic expansion has a greater influence on document matching, and the system will be more inclined to match documents with extended semantics.

[0133] 5. n is the detail complexity of the document, which measures the complexity of the detailed description of the document at the semantic level. A higher n value means that the document contains more detailed technical descriptions and experimental data.

[0134] Assume that the basic semantic matching function g1(s) of document x1 is 0.8, which means that the document contains a lot of basic content related to "artificial intelligence" and "medical care". The semantic extension function g2(s) of document x1 is 0.6, which means that there are many potentially relevant contents in the document, which may be combined with other studies. Assume that the semantic extension coefficient γ is 1.2, which means that the extended semantics has a strong influence on the document matching results. The detail complexity n of the document is 10, which means that the document contains a lot of detailed descriptions.

[0135] Substitute into the formula to calculate:

[0136]

[0137] L(x1)=1.52·0.316=0.48

[0138] Through this calculation, the semantic level score of document x1 is 0.48.

[0139] Similarly, assume that the basic semantic matching function g1(s) of document x2 is 0.6, the semantic expansion function g2(s) is 0.7, the detail complexity n is 8, and the semantic expansion coefficient γ is 1.0. Substitute into the formula for calculation:

[0140]

[0141] L(x2)=1.3·0.353=0.46

[0142] The semantic level score of document x2 is 0.46.

[0143] Assume that the basic semantic matching function g1(s) of document x3 is 0.7, the semantic expansion function g2(s) is 0.4, the detail complexity n is 12, and the semantic expansion coefficient γ is 0.9. Substitute into the formula for calculation:

[0144]

[0145] L(x3)=1.06·0.288=0.31

[0146] The semantic level score of document x3 is 0.31.

[0147] According to the calculation results, the semantic level score of document x1 is 0.48, document x2 is 0.46, and document x3 is 0.31. The system will dynamically adjust the display order of the documents based on these scores. When the user queries "specific applications of artificial intelligence in medical diagnosis and treatment", documents x1 and x2 will be displayed first because they have higher semantic matching scores, especially with the support of semantic expansion, these documents can provide richer and more relevant content.

[0148] Through semantic association analysis technology, the system can more intelligently understand the intention behind the user's query and perform deep matching based on the basic semantics and extended semantics of the document. Document x1 occupies a higher position in the search results because of its rich content and high semantic matching. Although document x3 has strong basic semantics, it has weak extended semantics and high detail complexity, so its final score is low and it is ranked at the back of the results.

[0149] Embodiment 2:

[0150] Combined with Figure 3 Flowchart, a user visits documents related to "machine learning" and "NLP" (natural language processing) multiple times over a period of time. The system uses logs to track and record each query behavior. For example, when a user searches for "machine learning applied to NLP", he visits documents x1, x2, and x3 respectively. The system incorporates these query behavior data into the log and keeps track of the user's subsequent behavior through cookie technology.

[0151] The system records the following behavior data:

[0152] Document x1: The user visited 8 times, and each visit lasted 15 minutes.

[0153] Document x2: The user visited 5 times, and stayed for 10 minutes each time.

[0154] Document x3: The user visited the document 3 times, and stayed for 20 minutes each time.

[0155] In order to analyze the user's search behavior, the system substitutes these behavior data into the following formula for calculation:

[0156]

[0157] in:

[0158] B(x) represents the comprehensive score of user behavior, which measures the user's interest in a certain document.

[0159] f1(x) is the access frequency function, which indicates the number of times document x is accessed by users.

[0160] f2(x) is the residence time function, which reflects the time the user stays on the document.

[0161] ω1 and ω2 are weight factors of access frequency and residence time, respectively. It is assumed that the system sets the weights of access frequency and residence time to ω1=0.7 and ω2=0.9, respectively.

[0162] ψ(t) is a time-dynamic adjustment function. Assume that the initial value of ψ(t) is 1.05 and gradually increases over time, indicating that the system gives higher priority to recent behaviors.

[0163] T is the behavior time window, assumed to be 30 days.

[0164] Substitute actual data for calculation:

[0165] 1. Document x1:

[0166] Visit frequency: 8 times.

[0167] Duration of each stay: 15 minutes.

[0168] Weight factors: ω1=0.7, ω2=0.9.

[0169] Dynamic adjustment coefficient: ψ(t)=1.05.

[0170] Substituting these data into the formula:

[0171]

[0172] The calculations show that:

[0173]

[0174] B(x1)=19.775×30=593.25

[0175] 2. Literature x2:

[0176] Visit frequency: 5 times.

[0177] Duration of each session: 10 minutes.

[0178] Weight factors: ω1=0.7, ω2=0.9.

[0179] Dynamic adjustment coefficient: ψ(t)=1.05.

[0180] Substituting the data into the formula:

[0181]

[0182] B(x2)=12.95×30=388.5

[0183] 3. Literature x3:

[0184] Visit frequency: 3 times.

[0185] Duration of each stay: 20 minutes.

[0186] Weight factors: ω1=0.7, ω2=0.9.

[0187] Dynamic adjustment coefficient: ψ(t)=1.05.

[0188] Substituting the data into the formula:

[0189]

[0190] B(x3)=21×30=630

[0191] According to the above calculation results, document x3 has the highest comprehensive score of 630, which means that although users visit x3 less frequently, the system considers the document to be highly relevant to users due to the longer stay time. Document x1 has a score of 593.25, indicating that the document is also highly attractive to users, mainly reflected in the high frequency of users' visits. Document x2 has a score of 388.5, which is relatively low, reflecting that the document is slightly less attractive to users.

[0192] Through this multi-dimensional modeling and calculation, the system can dynamically adjust the priority of documents in the index. When the user queries again for content related to "Application of machine learning in natural language processing", document x3 will be displayed first because it has the highest user behavior comprehensive score, although its access frequency is low. Document x1 will also be displayed with a higher priority because of its high access frequency.

[0193] In the previous scenario, the user has searched for literature related to "Application of Machine Learning in Natural Language Processing" many times, and the system has recorded the user's query behavior data through log tracking and Cookie technology. This data includes the category of documents visited by the user, the frequency of access, the query type, and the access path. Now, the system needs to further analyze this data, use the clustering analysis algorithm to identify the user's typical behavior pattern, and then use the recurrent neural network (RNN) model to predict the user's future retrieval needs.

[0194] First, the system uses KMeans or DBSCAN clustering algorithms to cluster the user's historical behavior data. Suppose through analysis, the system identifies two typical user usage patterns:

[0195] Mode A: Frequently search for literature related to "Basics of Machine Learning" and "Applications of NLP". The user stops at literature x1, x2, and x3 many times, and the path is relatively fixed.

[0196] Mode B: Cross-domain queries are often conducted, and the query topics extend from "deep learning" to "neural networks", with relatively scattered paths.

[0197] The cluster analysis results show that the user belongs to mode A, that is, the user mainly focuses on the literature on machine learning and natural language processing, which means that the user is likely to continue to query content related to these topics in future searches.

[0198] Based on the results of cluster analysis, the system then uses a recurrent neural network (RNN) model to predict the user's future retrieval needs. RNN is a deep learning model for processing sequence data that can learn and capture the temporal dependencies in user behavior sequences. The model predicts the user's future retrieval needs using the following formula:

[0199]

[0200] Among them, P(u) is the prediction of the user's future search needs, u i represents the ith retrieval behavior, f3(u i ) is the user behavior sequence feature function, describing the user's past retrieval behavior pattern, and λ is the adjustment parameter that controls the prediction sensitivity, ranging from 0.1 to 0.5, indicating the sensitivity of the model to recent behaviors. i Indicates the time when the behavior occurs, in days.

[0201] Assume that the user's search behavior in the past 30 days is:

[0202] Day 1: Search for literature related to "Basics of Machine Learning", behavior u1.

[0203] Day 7: Query "Deep Learning Applications in NLP", behavior u2.

[0204] Day 15: Query "Application of machine learning in healthcare", behavior u3.

[0205] Day 20: Query "Recent Advances in Natural Language Processing", behavior u4.

[0206] The corresponding behavior sequence characteristic function f3(u i) can be expressed as features related to document content and query path. Assume f3(u1) = 0.8, f3(u2) = 0.7, f3(u3) = 0.9, f3(u4) = 0.85, and set λ = 0.3.

[0207] We substitute these data into the formula to predict the user's future search needs:

[0208] 1. Day 1 behavior u1:

[0209]

[0210] 2. Day 7 behavior u2:

[0211]

[0212] 3. Day 15 behavior u3:

[0213]

[0214] 4. Day 20 behavior u4:

[0215]

[0216] According to these calculation results, the user's recent behavior u1 has the greatest impact on future retrieval needs, followed by behavior u2, while the more distant behaviors u3 and u4 have less impact on the prediction. From this, the system can conclude that the user's future retrieval topics will still focus on "Machine Learning Basics" and "Deep Learning Applications in NLP" related content.

[0217] Based on the prediction of the RNN model, the system can give priority to recommending documents related to "Basics of Machine Learning" and "Applications of NLP" to users when they re-enter the library system, especially those documents x1 and x2 that are most relevant to previous behaviors. This not only improves the accuracy of retrieval, but also saves users a lot of time by avoiding repeated searches for highly relevant content that has been queried before. At the same time, the system will also recommend a small amount of extended documents to users, such as the latest documents related to "deep learning" or "neural networks", to meet users' potential cross-domain interests.

[0218] By identifying user behavior patterns through cluster analysis and using the RNN model to predict users' future retrieval needs, the system can dynamically adjust indexing and recommendation strategies to provide users with more accurate literature recommendations.

[0219] In the above scenario, based on the calculation results of the recurrent neural network (RNN) prediction model, the system has predicted that the user's future search needs will focus on the literature related to "Machine Learning Basics" and "Applications in NLP", especially literature x1 and x2. In order to optimize the user's search experience, the system will load the index data of these related literature in advance to ensure that users can instantly obtain highly relevant literature when searching. At the same time, the system will also dynamically optimize the index order through the user's actual behavior feedback (such as query frequency and path click rate) to further improve the search efficiency.

[0220] According to the previously calculated prediction result P(u), documents x1 and x2 are predicted to be the documents that users are most likely to retrieve in the future. Therefore, the system will use an adaptive preloading mechanism to load the indexes of these documents in advance. For example, if the comprehensive score of document x1 is high (such as a score of 593.25 in the behavioral model), the system will prioritize loading the index of this document. Document x2 has a slightly lower score (such as 388.5) and will also be preloaded at the same time. Through this mechanism, the system can quickly respond to user query requests and reduce loading delays.

[0221] When the user begins to perform actual search behavior, the system will dynamically optimize the index sorting based on the user's query frequency and path click rate. The optimization process is calculated using the following formula:

[0222]

[0223] in:

[0224] S(y) represents the priority ranking score of the document in the index.

[0225] r1(y i ) is the query frequency function, indicating that the user queries document y i The number of times.

[0226] r2(y i ) is the path click rate function, which indicates the frequency of users clicking on the document.

[0227] γ1 and γ2 are adjustment coefficients, which control the impact of query frequency and path click rate on document priority. Assume that γ1 = 0.6 and γ2 = 0.8.

[0228] Assume that in the next week, the user searches for documents x1, x2, and x3 multiple times:

[0229] Document x1: query frequency r1(y1)=10, path click rate r2(y1)=8.

[0230] Document x2: query frequency r1(y2)=7, path click rate r2(y2)=5.

[0231] Document x3: query frequency r1(y3)=4, path click rate r2(y3)=3.

[0232] Based on these data, the system calculates a priority ranking score for each document:

[0233] 1. Document x1:

[0234]

[0235] 2. Literature x2:

[0236]

[0237] 3. Literature x3:

[0238]

[0239] According to the above calculation results, the system knows that document x1 has the highest priority ranking score of 12.4; document x2 has a score of 8.2; and document x3 has a score of 4.8. Therefore, the system will dynamically adjust the index order. The next time the user searches for a related topic, document x1 will be displayed first, followed by document x2, and document x3 will be ranked later. Through this dynamic optimization process, the system can ensure that documents with high query frequency and high path click rate have a higher priority in the index, improving the relevance and efficiency of document retrieval.

[0240] In this example, the system preloads the document index that best matches the user's needs based on the prediction results, and dynamically adjusts the document priority ranking based on the user's actual search behavior feedback. Document x1 maintains a higher priority in the index because of its high query frequency and click-through rate, while documents x2 and x3 have relatively lower priorities.

[0241] Embodiment 3:

[0242] Combined with Figure 3 Flowchart, suppose a user searches for "Application of deep learning in natural language processing" in the library. The system needs to identify the two core keywords "deep learning" and "natural language processing" and understand the user's specific needs based on the context of the query. At this time, the system uses the natural language processing model of deep learning to parse the query and calculates the context semantic vector of the query according to the formula.

[0243] The core formula of semantic parsing is:

[0244]

[0245] Where V(q) represents the contextual semantic vector of the query statement, which is used to quantify the semantic content of the user query; f1(w i ) is the keyword semantic weight function, which represents the i-th keyword w in the query i The weight in the semantic space; f2(c i ) is the context-dependent function, indicating that keyword w i With its context c i α and β are adjustment coefficients, which control the weights of keywords and context respectively. Assume that α = 0.7, which means that the system has a higher weight on keywords, and β = 0.6, which means that the context is less important to semantics.

[0246] When a user queries "Application of deep learning in natural language processing", it is assumed that the system identifies keywords w1 as "deep learning" and w2 as "natural language processing". Through the natural language processing model, the system performs semantic weight analysis on these keywords:

[0247] 1. Keyword w1 = "deep learning": System analysis shows that its semantic weight f1(w1) = 0.85, because the word highly matches the query topic.

[0248] 2. Keyword w2 = "natural language processing": semantic weight f1(w2) = 0.9, which is also the core word in the query topic.

[0249] 3. Context association function f2(c1): Assuming that "deep learning" is related to words such as "application" and "algorithm" in the context of the query, the given context association value is f2(c1)=0.75.

[0250] 4. Context association function f2(c2): For “natural language processing”, context words such as “technology”, “data”, etc. are identified as highly relevant words, f2(2)=0.8.

[0251] In addition, the system sets a time dynamic adjustment function γ(t) that changes over time, increasing from a range of 1.0 to 1.05, indicating that the system gradually adjusts the priority of query semantics according to the time when the query occurs.

[0252] Substituting these data into the formula for calculation, first calculate the semantic vectors of keywords w1 and w2:

[0253] 1. Calculation of semantic vector of keyword w1:

[0254]

[0255]

[0256] Assuming that the time window T for semantic analysis is 30 days, then:

[0257] V(w1)=1.0675×30=32.025

[0258] 2. Calculation of semantic vector of keyword w2:

[0259]

[0260] Assuming the time window is 30 days, then:

[0261] V(w2)=1.134×30=34.02

[0262] Through the above calculations, the system obtains the semantic vector V(w1) = 32.025 for the keyword "deep learning" and the semantic vector V(w2) = 34.02 for "natural language processing". These vectors will be used to measure the semantic match between the query and related documents, helping the system to better understand the user's query intent. The system will give priority to displaying documents that have a high relevance to these keywords, and combine the context for more accurate matching.

[0263] Based on the user's query semantic vector, the system can recognize that the user's focus on "the specific application of deep learning in natural language processing" is not only the basic principles of the technology, but also the specific implementation of the technology in practical applications. Since "natural language processing" and "deep learning" are the core of the query and are highly related to the "application" in the context, the system will give priority to displaying research results and application cases in related fields. For example, literature related to "deep learning algorithms" or papers related to "natural language processing applications" will be ranked at the top of the search results due to high semantic matching.

[0264] By using a deep learning natural language processing model, the system can not only identify the core keywords in user queries, but also understand the semantic intentions behind the user through context association. By calculating the contextual semantic vector of the query, the system can understand the user's needs more accurately and dynamically adjust the retrieval results of the document based on the query semantics.

[0265] Assuming that the user continues to search for "Application of deep learning in natural language processing" in the digital library, the system needs to not only identify the keywords in the query, but also distinguish the user's query type through semantic intent recognition technology. At this time, the user's query may belong to "information query", that is, the user wants to understand the specific information or technical principles of a certain field, or it may belong to "transaction query", that is, the user wants to find the specific implementation or application of a certain technology. In order to better display the documents that meet the user's needs, the system will use the intent classification model to hierarchically process the query and calculate the priority of the document through a formula.

[0266] The system first uses semantic intent recognition technology to classify the user's query "Application of deep learning in natural language processing" as an "information query" because the user may mainly want to obtain information about "the basic principles and applications of deep learning in NLP." At the same time, if the query contains clear goals or application scenarios, it may be classified as a "transaction query", such as "how to use deep learning models to process NLP data."

[0267] In order to build a multi-level index based on the user's query intent, the system uses the following formula to calculate the hierarchical score of the intent:

[0268]

[0269] Among them, I(s) is the hierarchical score of the user's query intention, which quantifies the priority of the user's query intention; g i (x) is the score of document x at different intent levels, reflecting the semantic match between the document and the user’s query. The system will i And the time sensitivity parameter λ is used to adjust the query priority. Assume that λ ranges from 0.1 to 0.5, which determines how fast the query intent changes over time.

[0270] To better understand this process, assume that the user's query behavior occurred in the past 30 days:

[0271] Day 5: Query “Basic Concepts of Deep Learning” (Information Query).

[0272] Day 10: Query “Common Algorithms in Natural Language Processing” (Information Query).

[0273] Day 20: Query “How to apply deep learning models in NLP” (Transactional Query).

[0274] For these queries, the system will calculate the scores of relevant documents at different intent levels. Suppose document x1 is about "deep learning foundation", document x2 is about "natural language processing algorithm", and document x3 is about "deep learning in NLP application". We set:

[0275] The semantic level score of document x1 is g1(x1)=0.9.

[0276] The semantic level score of document x2 is g1(x2)=0.85.

[0277] The semantic level score of document x3 is g1(x3)=0.95.

[0278] The system calculates different query behaviors of users, assuming that the time sensitivity λ = 0.3:

[0279] 1. Day 5 query "Basic concepts of deep learning":

[0280]

[0281] 2. Day 10 query "Common algorithms in natural language processing":

[0282]

[0283] 3. Day 20 query “How to apply deep learning models in NLP”:

[0284]

[0285] Through the above calculation, document x1 is the most matched with the user's "information query" and the time is relatively recent, so its hierarchical score is relatively high, about 0.1644. Therefore, when the user makes a related query again, this document will be displayed first. The hierarchical score of document x2 is 0.0403. Although it is highly relevant, it has a lower display priority because it has been a long time. Document x3 is a matching document for transaction query, with a hierarchical score of 0.0024, indicating that it has the lowest priority and may be ranked at the back.

[0286] By building a multi-level index, the system can dynamically adjust the priority of documents according to the user's query intent and query time, ensuring that the results obtained by the user are more in line with their current needs. For example, when a user has recently searched for information documents many times, the system will prioritize the display of relevant basic documents rather than application or transaction documents.

[0287] By using the intent classification model, the system can build a multi-level index based on the user's query type (information query, navigation query or transaction query) and dynamically adjust the priority display of documents. By calculating the scores of documents at different levels, the system can more accurately understand the user's intent and provide the documents that best match their needs.

[0288] When the user searches for "Application of deep learning in natural language processing" again in the digital library, the system should not only display directly matching documents, but also find other documents related to the topic but not directly containing the keywords "deep learning" or "natural language processing" through the semantic expansion mechanism. For example, the document may discuss "the role of neural networks in language models" or "the application of machine learning in text processing", which are semantically related to the query but do not use precise keywords.

[0289] In order to optimize the search results through the semantic expansion mechanism, the system uses the following formula to calculate the expanded recommended result set:

[0290]

[0291] in:

[0292] R(q) is the recommended search result set after semantic expansion.

[0293] h1(y i ) is a direct keyword matching result function, indicating that document y i The score of a direct match to query q.

[0294] h2(y i ) is the semantic expansion result function, which means document y i The relevance score to the query q at the semantic level.

[0295] μ is the semantic expansion coefficient, ranging from 0.5 to 2.0, which controls the influence of semantic expansion results on the final recommendation. Assume that μ = 1.2 here, it means that semantic expansion has a greater influence on the retrieval results.

[0296] n is the total number of retrieved documents.

[0297] Assume that when a user queries "Application of deep learning in natural language processing", the system identifies that the documents directly related to the query include y1, y2, and y3:

[0298] The keywords of document y1 completely match the query, and the direct score is h1(y1)=0.95.

[0299] Document y2 discusses “Application of Neural Networks in Language Processing”, which is semantically relevant to the query but has different keywords. Its expansion score h2(y2)=0.8.

[0300] Document y3 mainly discusses "the role of machine learning in text processing", which has a certain semantic relevance, and the expansion score h2(y3)=0.75.

[0301] According to the formula, the system combines the direct matching scores of these documents with the semantic expansion scores to calculate the expanded recommendation result set:

[0302] 1. The expanded recommendation score of document y1:

[0303]

[0304] Assuming the total number of documents n = 100, then:

[0305]

[0306] 2. The expanded recommendation score of document y2:

[0307]

[0308] 3. The extended recommendation score of document y3:

[0309]

[0310] Through calculation, the system found that although document y2 did not directly match the query keyword, due to its high semantic extension score, the final recommendation score R(y2) = 0.0096 exceeded the recommendation score R(y1) = 0.0095 of the directly matched document y1. The recommendation score of document y3 was 0.009, which was lower than the first two, but still had a high relevance. Therefore, when the user queries, the system will first recommend documents y2 and y1, and then recommend document y3.

[0311] Through this dynamic semantic expansion mechanism, the system not only displays documents that directly match the query keywords, but also mines documents that are semantically related to the query but not exactly matched. This greatly enriches the user's search results and ensures that users can find more valuable relevant documents. For example, although document y2 does not directly contain the keyword "deep learning", it discusses "the application of neural networks in natural language processing" in detail, which is equally important for users to understand the overall picture of this field.

[0312] In this example, the system expands the scope of search results based on the semantic intent of the user's query through a dynamic semantic expansion mechanism and knowledge graph. Documents are not only displayed in the search results through keyword matching, but also prioritized by semantic expansion scores. By adjusting the expansion coefficient μ and time characteristics, the system can ensure that the search results not only cover the user's direct needs, but also provide users with more potentially relevant content. This mechanism greatly improves the search breadth and depth of digital libraries, ensuring that users can obtain more comprehensive document support when conducting complex or multi-level queries.

[0313] Embodiment 4:

[0314] When the user searches for "Application of deep learning in natural language processing" again in the digital library, the system should not only display directly matching documents, but also find other documents related to the topic but not directly containing the keywords "deep learning" or "natural language processing" through the semantic expansion mechanism. For example, the document may discuss "the role of neural networks in language models" or "the application of machine learning in text processing", which are semantically related to the query but do not use precise keywords.

[0315] In order to optimize the search results through the semantic expansion mechanism, the system uses the following formula to calculate the expanded recommended result set:

[0316]

[0317] in:

[0318] R(q) is the recommended search result set after semantic expansion.

[0319] h1(y i ) is a direct keyword matching result function, indicating that document y i The score of a direct match to query q.

[0320] h2(y i ) is the semantic expansion result function, which means document y i The relevance score to the query q at the semantic level.

[0321] μ is the semantic expansion coefficient, ranging from 0.5 to 2.0, which controls the influence of semantic expansion results on the final recommendation. Assume that μ = 1.2 here, it means that semantic expansion has a greater influence on the retrieval results.

[0322] n is the total number of retrieved documents.

[0323] Assume that when a user queries "Application of deep learning in natural language processing", the system identifies that the documents directly related to the query include y1, y2, and y3:

[0324] The keywords of document y1 completely match the query, and the direct score is h1(y1)=0.95.

[0325] Document y2 discusses “Application of Neural Networks in Language Processing”, which is semantically relevant to the query but has different keywords. Its expansion score h2(y2)=0.8.

[0326] Document y3 mainly discusses "the role of machine learning in text processing", which has a certain semantic relevance, and the expansion score h2(y3)=0.75.

[0327] According to the formula, the system combines the direct matching scores of these documents with the semantic expansion scores to calculate the expanded recommendation result set:

[0328] 1. The expanded recommendation score of document y1:

[0329]

[0330] Assuming the total number of documents n = 100, then:

[0331]

[0332] 2. The expanded recommendation score of document y2:

[0333]

[0334] 3. The extended recommendation score of document y3:

[0335]

[0336] Through calculation, the system found that although document y2 did not directly match the query keyword, due to its high semantic extension score, the final recommendation score R(y2) = 0.0096 exceeded the recommendation score R(y1) = 0.0095 of the directly matched document y1. The recommendation score of document y3 was 0.009, which was lower than the first two, but still had a high relevance. Therefore, when the user queries, the system will first recommend documents y2 and y1, and then recommend document y3.

[0337] Through this dynamic semantic expansion mechanism, the system not only displays documents that directly match the query keywords, but also mines documents that are semantically related to the query but not exactly matched. This greatly enriches the user's search results and ensures that users can find more valuable relevant documents. For example, although document y2 does not directly contain the keyword "deep learning", it discusses "the application of neural networks in natural language processing" in detail, which is equally important for users to understand the overall picture of this field.

[0338] In this example, the system expands the scope of search results based on the semantic intent of the user's query through a dynamic semantic expansion mechanism and knowledge graph. Documents are not only displayed in the search results through keyword matching, but also prioritized by semantic expansion scores. By adjusting the expansion coefficient μ and time characteristics, the system can ensure that the search results not only cover the user's direct needs, but also provide users with more potentially relevant content.

[0339] Embodiment 5:

[0340] Combined with Figure 5 Flowchart, suppose a user searches for "Application of Deep Learning in Natural Language Processing" in a digital library. The system adjusts the semantic priority and index structure of documents based on the user's click behavior and dwell time in the search results. The user clicks on three documents v1, v2, and v3, and dwells on each document for different periods of time.

[0341] The system first collects the user's click behavior and dwell time in the search results, and uses the formula to calculate the semantic priority score of the document. The formula is as follows:

[0342]

[0343] in:

[0344] P(u) represents the semantic priority score of document u.

[0345] C u(t) is the change of the click rate of the document over time, indicating the user's click behavior at time t.

[0346] S u (t) is the residence time function of the document, reflecting the time the user stays on the document page.

[0347] α = 0.6 and β = 0.8 are coefficients that control the importance of click rate and dwell time, and λ = 0.05 is the time decay factor.

[0348] Assume that in the past 30 days, the user's behavior data for documents v1, v2, and v3 are as follows:

[0349] Literature v1: Click-through rate Dwell time minute.

[0350] Literature v2: Click-through rate Dwell time minute.

[0351] Literature v3: Click-through rate Dwell time minute.

[0352] Substitute these data into the formula to calculate the semantic priority score of the document.

[0353] 1. Semantic priority of document v1:

[0354]

[0355] Assume t = 30 days:

[0356] P(v1)≈0.6·2.77+0.8·10·e -1.5 ≈1.662+0.8·10·0.223≈1.662+1.784=3.446

[0357] 2. Semantic priority of document v2:

[0358]

[0359] The calculations show that:

[0360] P(v2)≈0.6·2.2+0.8·5·0.223≈1.32+0.892=2.212

[0361] 3. Semantic priority of document v3:

[0362]

[0363] The calculations show that:

[0364] P(v3)≈0.6·2.56+0.8·7·0.223≈1.536+1.249=2.785

[0365] S2: Dynamically adjust the semantic model and index structure of documents

[0366] The system inputs these behavioral feedbacks into the semantic matching model, uses deep learning algorithms to analyze long-term behaviors, and optimizes the position of documents in the semantic space. At the same time, the system uses a reinforcement learning algorithm to dynamically adjust the indexing priority of documents based on the dwell time and click rate. The formula is as follows:

[0367]

[0368] in:

[0369] I(v) is the index priority score of document v.

[0370] S i (v) and C i (v) are the dwell time and click rate during the i-th search, respectively.

[0371] μ=0.4 is the coefficient of the influence of control feedback on priority adjustment.

[0372] Based on the previous data, the indexing priority of documents v1, v2, and v3 is calculated as follows:

[0373] 1. Index priority of document v1:

[0374]

[0375] 2. Indexing priority of document v2:

[0376]

[0377] 3. Indexing priority of document v3:

[0378]

[0379] Combined with user behavior feedback, the system will also dynamically optimize the semantic expansion model. According to the user's click and stay time, the system dynamically optimizes and expands the keyword pool. The formula is as follows:

[0380]

[0381] in:

[0382] E(q) is the semantic expansion score of query q.

[0383] f1(w i ) is the keyword w iThe expansion weight, f2(c i ) is the context semantic expansion function.

[0384] k1=0.7 and k2=0.5 are expansion parameters that control the relative influence of keywords and context.

[0385] t is the time when the query occurs.

[0386] Assume that the user queries "Application of deep learning in natural language processing" and is related to the keywords "neural network" and "text processing". After the semantic extension model is optimized, the system identifies the relevant documents y1 and y2 and calculates their extension scores.

[0387] 1. Document y1:

[0388]

[0389] 2. Document y2:

[0390]

[0391] The system dynamically adjusts the order of document display by optimizing the semantic priority and index structure of documents through behavioral feedback. For example, documents v1 and v3 have a higher index priority and will be displayed first in future search results. In addition, the system uses a semantic expansion mechanism to identify documents y1 and y2 that the user may be interested in but are not directly matched, and recommends these documents to the user. This process ensures that users get a more personalized and accurate search experience.

[0392] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An optimization method for index retrieval of a digital library, characterized in that include: S1. Dynamic hierarchical index construction technology: S1.

1. First, conduct a multi-dimensional content analysis of the digital resources in the library, including subject, keyword, author, and publication time information; classify the documents based on the information and construct a content hierarchy; S1.

2. Based on the classification, dynamic indexes are constructed for different document levels including topics, content details, and semantic associations, so that commonly used high-frequency content can be quickly retrieved, while detailed content is retained in the deep index; S1.

3. Regularly adjust the index structure based on user usage frequency and search habits to keep frequently used information close to the surface; S2. Index optimization strategy based on usage pattern: S2.

1. By analyzing the user's search behavior, monitoring the user's access frequency, query type and access path in different categories of documents, and identifying typical user usage patterns; S2.

2. Dynamically adjust the priorities of different index items based on user behavior monitoring results, and promote frequently accessed documents or related topics to the top of the search results; S2.

3. Cache the high-frequency index parts locally according to the usage pattern; S3, semantic association retrieval optimization: S3.

1. Introduce natural language processing technology to perform semantic analysis on user query statements, identify keywords and understand the semantic intent behind them, and expand the index scope; S3.2, based on semantic analysis, intelligently match the user's query with the semantics of the document content, optimize the index, and enable documents with similar semantics but different keywords to be retrieved; S3.

3. Optimize the semantic matching model and index structure based on the click-through rate of search results and user dwell time feedback; S4. Index update and resource allocation optimization: S4.

1. Update the index in real time when new literature resources are added to the library or existing resources are modified, and adjust the index structure according to content classification and user behavior; S4.2, using distributed database technology to store indexes on multiple nodes and support large-scale parallel retrieval; S4.

3. During the retrieval process, resources are dynamically allocated and each retrieval node is scheduled according to the current query volume and load conditions.

2. The method for optimizing index retrieval of a digital library according to claim 1, characterized in that The method for constructing the dynamic index includes: S1. First, use content analysis and classification algorithms to conduct multi-dimensional analysis of digital resources in the library, including themes, content details, and semantic association information, and establish a preliminary structure of a dynamic index based on the document level; S2. Secondly, by monitoring user behavior, including query frequency, access time, and user path, the frequency-aware model is used to dynamically adjust the index position of the document; S3. Finally, through the semantic association analysis technology, deep semantic matching of documents is carried out. The semantic vector model based on deep learning is used to dynamically optimize the semantic level matching of different document contents. The semantic level is determined by the formula: Among them, L(s) represents the semantic level score of the document, which is the result of dividing the document into levels through semantic analysis; g1(s) is the basic semantic matching function, which represents the basic semantic association of the document; g2(s) is the semantic extension function, which is responsible for matching the document in the semantic space; γ is the semantic extension coefficient, which controls the impact of semantic extension on the matching results; n is the detail complexity of the document, which is used to measure the complexity of the detailed description of the document at the semantic level.

3. The method for optimizing index retrieval of a digital library according to claim 1, characterized in that The method for identifying a typical user usage pattern includes: S1. First, log tracking and Cookie technology are used to record the user's search behavior data in real time, including the document category, query type, access path and stay time of each visit; S2. Then, based on the collected behavioral data, cluster analysis algorithms including K-Means or DBSCAN are used to classify users, identify typical user usage patterns, including users who frequently query specific categories of documents or follow similar paths, and mathematically model the patterns.

4. The method for optimizing index retrieval of a digital library according to claim 3 is characterized in that Based on the prediction results of P(u), the index of relevant documents is loaded in advance, and the retrieval efficiency is optimized by using an adaptive preloading mechanism; then, according to the actual feedback of user behavior, the display order of the index is adjusted, and the dynamic sorting of the index is optimized by using the formula: Among them, S(y) represents the priority ranking score of the document in the index; r1(y i ) and r2(y i ) are the user's query frequency and path click rate function in the document, γ1 and γ2 are adjustment coefficients, which control the influence of the two on the priority.

5. The method for optimizing index retrieval of a digital library according to claim 1, characterized in that The framework used to identify keywords and understand the semantic intent behind them is as follows: S1. First, use the deep learning natural language processing model to perform deep semantic analysis on the user query, and use the following formula to calculate the context semantic vector of the user query: Where V(q) represents the contextual semantic vector of the query statement, which is used to quantify the semantic content of the user query; f1(w i ) is the keyword semantic weight function, which represents the i-th keyword w in the query i The weight in the semantic space reflects the relevance of the keyword to the query topic; f2(c i ) is the context association function, indicating that the keyword w i With its context c i The semantic relationship between them; α and β are adjustment coefficients, which are used to balance the weights of keywords and context; γ(t) is a time dynamic adjustment function, which represents the semantic characteristics of the query changing over time; T is the time window of semantic parsing, which represents the duration of semantic parsing; S2. Then, we use semantic intent recognition technology to classify user queries into different types through the intent classification model, including information queries, navigation queries, and transaction queries, and use the formula to construct a multi-level index for the intent: Among them, I(s) represents the hierarchical score of query intent, which is an indicator used to quantify user query intent; g i (x) is a semantic level function, which indicates the score of document x in different intent levels and reflects the semantic matching degree between document and user query; λ is a parameter that controls the time sensitivity of the model and determines the speed at which query intent changes over time; t i is the timestamp of the query; S3. Finally, through the dynamic semantic expansion mechanism, the knowledge graph or deep semantic model is used to expand keywords according to the semantic intent of the user's current query statement. The calculation formula is: Among them, R(q) is the recommended search result set after expansion, which means the final search result after semantic expansion; h1(y i ) is the direct keyword matching result function, which represents the document y directly related to the query q i score; h2(y i ) is a semantic expansion result function, which represents documents y that are semantically related to query q but not directly keyword matched. i score; μ is the semantic expansion coefficient, which controls the influence of the expansion result on the final recommendation; n is the total number of retrieved documents.

6. The method for optimizing index retrieval of a digital library according to claim 1, characterized in that The optimization of the semantic matching model and index structure includes the following steps: S1. First, collect the user's click behavior and dwell time in the search results in real time, associate the data with the semantic relevance of the document, and calculate the impact of user behavior feedback on the semantic priority of the document based on the formula: Where P(u) represents the semantic priority score of document u, C u (t) is the change of the click rate of the document over time t, S u (t) is the user residence time function of the document, α and β are adjustment coefficients that control the relative importance of click-through rate and residence time, λ is the time decay factor, and T is the collection time period of feedback data; S2. Then, the behavior feedback is input into the semantic matching model to dynamically adjust the position of the document in the semantic space. The deep learning algorithm is used to analyze the long-term behavior and optimize the parameters of the semantic model. Then, based on the user's stay time and click-through rate, the reinforcement learning algorithm is used to dynamically adjust the index structure and update the index priority of the document through the formula: Among them, I(v) is the priority score of document v in the index, S i (v) is the residence time of the document at the time of the i-th search, C i (v) is the i-th click rate, μ is the coefficient of the control feedback on the priority adjustment, t i is the timestamp of the action; S3. Finally, we optimize the semantic expansion model based on behavioral feedback and dynamically optimize and expand the keyword pool based on user clicks and dwell time.

Images