Artificial intelligence-based scientific instrument procurement selection method and device

Abstract

The application provides a scientific instrument procurement selection method and device based on artificial intelligence, and belongs to the technical field of artificial intelligence, and comprises the following steps: constructing a knowledge graph of all initial instruments; calculating the evaluation value of each initial instrument in each evaluation dimension; matching in the knowledge graph according to the selection requirement information input by a user to obtain a candidate instrument set; extracting the evaluation value of each candidate instrument in the candidate instrument set in all evaluation dimensions; performing weighted calculation on the evaluation values corresponding to all candidate instruments according to the dimension weight parameters set by the user to obtain a personalized comprehensive score; and generating a selection scheme containing an instrument recommendation result and data traceability information according to the personalized comprehensive score and the associated information in the knowledge graph. The application constructs a knowledge graph fusing multi-source data, and generates a scientific instrument procurement selection scheme based on artificial intelligence in combination with multi-dimensional evaluation, so that accurate recommendation of scientific instrument procurement selection based on artificial intelligence is realized.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method and apparatus for procuring and selecting scientific instruments based on artificial intelligence. Background Technology

[0002] In scientific instrument analysis, the selection and procurement of instruments can significantly impact the accuracy of test results. The current market offers a wide variety of instruments with complex performance parameters, and significant differences exist between different manufacturers and models. Users often face multiple challenges during the selection process: on the one hand, instrument prices are generally high, making it difficult to balance budget constraints with performance requirements; on the other hand, the diverse functions of various instruments make it crucial to consider whether key parameters align with actual testing needs. Furthermore, the abundance of brands and varying user reviews further complicate the decision-making process. Therefore, users often need to comprehensively weigh factors such as performance, price, and user reputation.

[0003] However, existing selection methods mainly rely on human experience or recommendation algorithms based on historical data. Human experience suffers from strong subjectivity, scattered information sources, and limited comparison dimensions, while recommendation algorithms based on historical data rely heavily on past purchase records and user preferences, lacking dynamic comparison of instrument performance indicators with industry standard requirements, and failing to incorporate real-time market prices and the latest user feedback for updates, resulting in low accuracy of recommendation results. Summary of the Invention

[0004] This invention provides a method and apparatus for purchasing and selecting scientific instruments based on artificial intelligence, in order to overcome the deficiencies in the prior art and thereby improve the accuracy of recommendations for purchasing and selecting scientific instruments based on artificial intelligence.

[0005] This invention provides a scientific instrument procurement and selection method based on artificial intelligence, comprising the following steps:

[0006] A knowledge graph is constructed based on the relevant data from all initial instruments across various data sources.

[0007] Based on at least one of the relevant data and multiple preset evaluation dimensions, calculate the evaluation value of each initial instrument under each evaluation dimension;

[0008] The selection requirements input by the user are matched in the knowledge graph to obtain a set of candidate instruments;

[0009] Extract the evaluation value of each candidate instrument in the candidate instrument set under all the evaluation dimensions;

[0010] The evaluation values ​​of all candidate instruments are weighted according to the dimension weight parameters set by the user to obtain a personalized comprehensive score.

[0011] Based on the personalized comprehensive score and the association information in the knowledge graph, a selection scheme is generated that includes instrument recommendation results and data traceability information.

[0012] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein the relevant data includes standard information data, instrument evaluation index data, manufacturer instrument data, manufacturer instrument index data, winning bid data, user discussion data, and user relationship data.

[0013] The construction of a knowledge graph based on relevant data from all initial instruments across various data sources includes:

[0014] Based on the standard information data, the instrument measurement index data, the manufacturer instrument data, the manufacturer instrument index data, the winning bid data, the user discussion data, and the user relationship data, define the standard entity, the standard instrument entity, and the manufacturer instrument entity;

[0015] A first association is established between the standard instrument entity and the manufacturer instrument entity through instrument information and indicator names; a second association is established between the manufacturer instrument entity and the manufacturer instrument indicator data through instrument information; a third association is established between the manufacturer instrument entity and the winning bid data through instrument information; and a fourth association is established between user discussion data and user relationship data through user ID.

[0016] The knowledge graph is constructed based on the standard entity, the standard instrument entity, the manufacturer instrument entity, the first association, the second association, the third association, and the fourth association.

[0017] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein the standard entity includes all fields of the standard information data, as well as standard name vector values ​​and industry vector values; the standard instrument entity includes all fields of the instrument measurement index data, as well as instrument name vector values, application industry vector values, indicator item vector values, principle method vector values, and application scenario vector values; and the manufacturer instrument entity includes all fields of the manufacturer instrument data, as well as instrument name vector values, brand vector values, manufacturer name vector values, and place of origin vector values.

[0018] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein calculating the evaluation value of each initial instrument under each evaluation dimension based on at least one of the aforementioned relevant data and multiple preset evaluation dimensions includes:

[0019] Based on the manufacturer's instrument performance data and the instrument measurement data, calculate the functional performance evaluation value for each initial instrument;

[0020] Based on the manufacturer's instrument data and the winning bid data, calculate the price assessment value for each initial instrument;

[0021] Based on the user discussion data and the user relationship data, calculate the user usage evaluation value for each initial instrument.

[0022] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein calculating the functional index evaluation value of each initial instrument based on the manufacturer's instrument indicator data and the instrument measurement index data includes:

[0023] For each core indicator of the initial instrument, the manufacturer's claimed value in the manufacturer's instrument indicator data is compared with the benchmark value in the instrument measurement indicator data, and the difference assessment value is calculated according to the indicator type.

[0024] Calculate the mean value of the core indicators of the initial instrument based on the difference evaluation values ​​of all the core indicators;

[0025] Determine the percentage of the core indicators among all instrument indicators;

[0026] The evaluation value of the functional indicator is calculated based on the average value of the core indicator and the proportion of the quantity.

[0027] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein calculating the price assessment value of each initial instrument based on the manufacturer's instrument data and the winning bid data includes:

[0028] Valid winning bid records that match the initial instrument and are within a preset time range are selected from the winning bid data, and outliers that deviate from the average price of similar instruments by more than a preset standard deviation multiple are removed.

[0029] The average price of valid winning bids is calculated as the market average winning bid price for the initial instrument.

[0030] Extract the manufacturer's displayed price range for the initial instrument from the manufacturer's instrument data;

[0031] Calculate the manufacturer's quoted price based on the price range displayed by the manufacturer;

[0032] The difference between the manufacturer's quoted price and the average winning bid price in the market is used to obtain the winning bid difference value;

[0033] The initial instrument's bid fluctuation value is determined based on the difference between the maximum and minimum values ​​among all the valid bid-winning records.

[0034] The price assessment value is obtained based on the relationship between the difference in the winning bid and the fluctuation value of the winning bid.

[0035] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein calculating the user evaluation value of each initial instrument based on the user discussion data and the user relationship data includes:

[0036] A user attention network is constructed based on user relationship data, where users are nodes in the user attention network and the attention relationships between users are directed edges in the user attention network;

[0037] Based on the user attention network, the credibility value of each user is iteratively calculated;

[0038] Sentiment analysis is performed on each comment in the user discussion data to obtain a sentiment score;

[0039] For any of the aforementioned comment contents, the sentiment score is multiplied by the credibility value of the user who published the comment content to obtain a weighted score;

[0040] For any of the initial instruments, the weighted average of the scores of all the comments is calculated to obtain the user usage evaluation value.

[0041] According to the present invention, a scientific instrument procurement and selection method based on artificial intelligence is provided, wherein the step of matching the selection requirements information input by the user in the knowledge graph to obtain a candidate instrument set includes:

[0042] The selection requirements information is parsed using a large language model to extract at least one key piece of information, and corresponding key-value pairs are generated based on the key information.

[0043] Using the key-value pairs as query conditions, a retrieval is performed in the knowledge graph, wherein, for entities containing vector value attributes, the similarity between the vector of the corresponding field in the query conditions and the entity attribute vector in the knowledge graph is calculated;

[0044] Entities with similarity greater than a preset threshold are filtered out and sorted according to the association rules between entities and the user attention rules to form the candidate instrument set.

[0045] This invention also provides a scientific instrument procurement and selection device based on artificial intelligence, comprising the following modules:

[0046] The first processing module is used to construct a knowledge graph based on the relevant data from all initial instruments under various data sources;

[0047] The second processing module is used to calculate the evaluation value of each initial instrument under each evaluation dimension based on at least one of the relevant data and multiple preset evaluation dimensions.

[0048] The third processing module is used to match the selection requirements information input by the user in the knowledge graph to obtain a set of candidate instruments;

[0049] The fourth processing module is used to extract the evaluation values ​​of each candidate instrument in the candidate instrument set under all the evaluation dimensions;

[0050] The fifth processing module is used to perform weighted calculations on the evaluation values ​​corresponding to all the candidate instruments according to the dimension weight parameters set by the user, so as to obtain a personalized comprehensive score;

[0051] The sixth processing module is used to generate a selection scheme that includes instrument recommendation results and data traceability information based on the personalized comprehensive score and the association information in the knowledge graph.

[0052] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the artificial intelligence-based scientific instrument procurement and selection method as described above.

[0053] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the artificial intelligence-based scientific instrument procurement and selection method as described above.

[0054] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the artificial intelligence-based scientific instrument procurement and selection method as described above.

[0055] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0056] By constructing a knowledge graph based on relevant data from various data sources for all initial instruments, a structured association of different types of data is achieved, forming a multi-dimensional knowledge network covering instrument performance, price, and user usage. By calculating the evaluation value of each initial instrument across various evaluation dimensions based on at least one relevant data point, a quantitative comparison of instruments in performance, price, and user usage is achieved, ensuring the evaluation results reflect the true performance of the instruments. Matching user-input selection requirements within the knowledge graph ensures a precise correspondence between selection needs and instrument characteristics, guaranteeing the candidate instrument set includes only instruments that meet the user's actual testing scenarios and application requirements. By extracting the evaluation values ​​of each candidate instrument across all evaluation dimensions and weighting the evaluation values ​​according to user-defined dimension weight parameters, a highly relevant candidate set can be selected from a massive number of instruments, and the final recommendation ranking accurately reflects the user's actual purchasing preferences. Finally, by generating a selection scheme containing instrument recommendation results and data traceability information based on personalized comprehensive scores and association information in the knowledge graph, the objectivity and traceability of the selection output are achieved, thereby improving the accuracy and reliability of AI-based scientific instrument procurement and selection. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0058] Figure 1 This is one of the flowcharts illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention.

[0059] Figure 2 This is the second flowchart illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention.

[0060] Figure 3 This is the third flowchart illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention.

[0061] Figure 4 This is the fourth flowchart illustrating the scientific instrument procurement and selection method based on artificial intelligence provided by this invention.

[0062] Figure 5 This is the fifth flowchart illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention.

[0063] Figure 6This is the sixth flowchart illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention.

[0064] Figure 7 This is the seventh flowchart illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention.

[0065] Figure 8 This is a schematic diagram of the structure of the scientific instrument procurement and selection device based on artificial intelligence provided by the present invention.

[0066] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0068] It should be noted that in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a 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 elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The terms "upper," "lower," etc., indicating orientation or positional relationships according to the accompanying drawings, are only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the system or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0069] The terms "first," "second," etc., used in this invention are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0070] The following is combined Figures 1-9 This invention describes the method, apparatus, electronic device, storage medium, and computer program product for procuring and selecting scientific instruments based on artificial intelligence.

[0071] Reference Figure 1 , Figure 1 This is one of the flowcharts illustrating the artificial intelligence-based scientific instrument procurement and selection method provided by this invention, such as... Figure 1 As shown, steps 101 to 106 are included:

[0072] Step 101: Construct a knowledge graph based on the relevant data from all initial instruments across various data sources.

[0073] This invention provides an artificial intelligence-based method for scientific instrument procurement and selection. First, a knowledge graph is constructed based on relevant data from various data sources for all initial instruments. The instrument selection process involves a large amount of scattered and heterogeneous data, such as standard documents from different sources, product manuals provided by manufacturers, scattered bidding announcements, and unstructured user discussions. These data, with their inconsistent formats and complex content, make it difficult for selectors to comprehensively and efficiently obtain the information needed for decision-making, affecting the objectivity and accuracy of the selection. To solve these problems, it is necessary to first structurally integrate this multi-source heterogeneous data to form a unified and interconnected knowledge system.

[0074] The data sources may include, but are not limited to, publicly released standard databases, product information databases provided by instrument manufacturers, publicly released bidding announcement databases from government procurement platforms, user discussion databases in professional forums or user communities, and user social attention databases.

[0075] Relevant data refers to specific information extracted from the aforementioned data sources that is directly related to the instrument's performance indicators, market price, user reviews, applicable standards, etc. Specifically, relevant data includes standard information data, instrument measurement indicator data, manufacturer instrument data, manufacturer instrument indicator data, winning bid data, user discussion data, and user relationship data.

[0076] Knowledge graphs are a way of organizing and storing related data in a graph structure. They formally describe the concepts, entities and their semantic relationships in the real world that are related to the procurement and selection of scientific instruments based on artificial intelligence, forming an interconnected knowledge network.

[0077] In one specific implementation, a knowledge graph is constructed based on the relevant data of all initial instruments from various data sources. First, core concepts, such as instruments, standards, manufacturers, and users, are identified from the relevant data from different sources and defined as entities in the knowledge graph. Then, the relationships between these entities are analyzed and established; for example, a certain instrument conforms to a certain standard, a certain manufacturer produces a certain brand of instrument, and a certain user evaluates a certain model of instrument. By using the identified entities as nodes and the relationships between entities as edges, a network-like knowledge structure, i.e., the knowledge graph, is ultimately formed.

[0078] In one specific implementation, refer to Figure 2 , Figure 2 This is the second flowchart of the artificial intelligence-based scientific instrument procurement and selection method provided by the present invention. The specific process of constructing the knowledge graph includes steps 201 to 203:

[0079] Step 201: Define the standard entity, standard instrument entity, and manufacturer instrument entity based on the standard information data, instrument measurement index data, manufacturer instrument index data, winning bid data, user discussion data, and user relationship data.

[0080] The standard information data comes from publicly available documents such as international standards, national standards, or industry standards; the instrument measurement index data is the specific requirements for instrument performance derived from the aforementioned standard documents; and the manufacturer's instrument data, manufacturer's instrument index data, winning bid data, user discussion data, and user relationship data mainly come from our own platform and publicly available information.

[0081] The purpose of this step is to classify raw data from different sources into different core business entities based on their inherent attributes.

[0082] In one specific implementation, a standard entity is defined as all fields containing standard information data, such as standard number, standard title, and publication time, and additionally defines standard name vector values ​​and industry vector values ​​for semantic retrieval; a standard instrument entity is defined as all fields containing instrument measurement index data, such as instrument name, index item, and index value, and additionally defines instrument name vector values, application industry vector values, index item vector values, principle method vector values, and application scenario vector values; a manufacturer instrument entity is defined as all fields containing manufacturer instrument data, such as instrument model, price information, and place of origin, and additionally defines instrument name vector values, brand vector values, manufacturer name vector values, and place of origin vector values.

[0083] By vectorizing text information, unstructured text descriptions can be converted into numerical vectors that can be used for similarity calculations, laying the foundation for subsequent intelligent matching.

[0084] Step 202: Establish the first association between the standard instrument entity and the manufacturer's instrument entity through instrument information and indicator name; establish the second association between the manufacturer's instrument entity and the manufacturer's instrument indicator data through instrument information; establish the third association between the manufacturer's instrument entity and the winning bid data through instrument information; and establish the fourth association between user discussion data and user relationship data through user ID.

[0085] This step aims to clarify the logical connections between different entities, thereby linking isolated entities into a meaningful knowledge network.

[0086] In practice, the first association is established by matching the instrument information and indicator names in the instrument measurement index table with the corresponding fields in the manufacturer's instrument index table. This association clarifies whether a specific instrument corresponds to a certain standard requirement for a particular performance indicator. The second association connects a manufacturer's instrument entity with its multiple detailed indicator data records, ensuring a complete correspondence between an instrument and all its performance parameters. The third association connects a manufacturer's instrument entity with one or more winning bid records, thus linking the instrument to its actual market transaction performance. The fourth association connects user-posted discussion content with the user's social media following relationships using user IDs, providing a basis for subsequently assessing the credibility of user statements.

[0087] Step 203: Construct a knowledge graph based on standard entities, standard instrument entities, manufacturer instrument entities, first association, second association, third association, and fourth association.

[0088] Specifically, the aforementioned standard entities, standard instrument entities, and manufacturer instruments are used as nodes in the knowledge graph. The first, second, third, and fourth relationships between entities are used as edges connecting the nodes. All types of attribute information contained in the entities and relationships are stored in the graph database.

[0089] By performing the above steps, instrument-related data from different data sources and in various formats can be integrated and transformed into a structured, semantic knowledge graph. This knowledge graph not only stores comprehensive information needed for AI-based scientific instrument procurement and selection in a unified format, but more importantly, it clearly reveals the intrinsic connections between standards, instrument performance, market prices, and user reputation through pre-defined relationships. This network-like knowledge structure provides a solid data foundation for subsequent complex relational queries, multi-dimensional evaluations, and precise recommendations, effectively solving the technical problems of information silos and low data utilization efficiency faced by traditional methods.

[0090] After constructing the knowledge graph, the artificial intelligence-based scientific instrument procurement and selection method provided by this invention also includes:

[0091] Step 102: Calculate the evaluation value of each initial instrument under each evaluation dimension based on at least one relevant data.

[0092] In actual instrument procurement decisions, selectors need to comprehensively consider multiple aspects, such as whether the performance meets standards, whether the price is reasonable, and the actual user experience. However, this information is often qualitative, scattered, and lacks unified evaluation criteria, making it difficult for selectors to conduct objective horizontal comparisons. Therefore, to address the problem of strong subjectivity in existing instrument evaluation processes, this application establishes an evaluation method that can convert performance across different dimensions into quantifiable scores.

[0093] The initial instruments refer to all the instruments to be evaluated contained in the knowledge graph.

[0094] Evaluation dimensions refer to different aspects used to measure the overall value of an instrument, such as the instrument's core parameters and other functional indicators, the instrument's market price, and the actual usage by users.

[0095] Evaluation values ​​are numerical values ​​used to quantitatively characterize the performance of an instrument under a specific evaluation dimension.

[0096] Specifically, for different evaluation dimensions, one or more relevant data are selected as the basis for calculation. For example, when evaluating functional indicators such as core instrument parameters, the main data can be the manufacturer's instrument specifications and measurement indicators; when evaluating the instrument's market price, the main data can be the manufacturer's instrument data and winning bid data; and when evaluating user usage, the main data can be user discussion data and user relationship data. By analyzing and calculating these relevant data, a specific evaluation value is generated for each initial instrument under each evaluation dimension.

[0097] In one specific implementation, refer to Figure 3 , Figure 3 This is the third flowchart of the artificial intelligence-based scientific instrument procurement and selection method provided by the present invention. Step 102 specifically includes steps 301 to 303:

[0098] Step 301: Calculate the functional index evaluation value for each initial instrument based on the manufacturer's instrument specification data and the instrument measurement index data.

[0099] Specifically, this application first calculates the initial instrument's functional index evaluation value from the instrument's functional dimension.

[0100] The functionality and performance of an instrument are key factors in the selection decision, determining whether it can meet actual testing needs. However, in practical applications, the instrument specifications provided by manufacturers may be exaggerated or based on inconsistent data sources, and the benchmark values ​​specified in standard documents are constantly being updated. How to objectively evaluate whether an instrument's performance truly meets the standards and quantify its degree of compliance is a pressing technical problem that needs to be solved in the procurement and selection process of artificial intelligence-based scientific instruments.

[0101] In step 301, the manufacturer's instrument specification data refers to the instrument performance parameters that the manufacturer claims or indicates in the product manual, such as detection accuracy, mass range, and resolution. Instrument performance evaluation data refers to the performance benchmark requirements that the instrument should meet for a specific application scenario, extracted from standard documents. Functional performance evaluation values ​​are numerical values ​​used to quantify the quality of an instrument's functional performance; these values ​​typically range from 0 to 1, with values ​​closer to 1 indicating better functional performance.

[0102] In one specific implementation, the performance indicators of similar instruments with the same function are first identified and matched in manufacturer data and standard data. Then, based on the characteristics of the indicators and the reliability of the data, appropriate algorithms are used for calculation. The characteristics of the indicators refer to whether a larger or smaller value is better. Finally, by synthesizing the calculation results of all key indicators of the initial instrument, an evaluation value reflecting the overall functional performance level of the instrument is obtained.

[0103] In a preferred embodiment, refer to Figure 4 , Figure 4 This is the fourth flowchart of the artificial intelligence-based scientific instrument procurement and selection method provided by the present invention. The specific process of calculating the functional index evaluation value of each initial instrument includes steps 401 to 404:

[0104] Step 401: For each core indicator of the initial instrument, compare the manufacturer's claimed value in the manufacturer's instrument indicator data with the benchmark value in the instrument measurement indicator data, and calculate the difference assessment value according to the indicator type.

[0105] Step 402: Calculate the average value of the core indicators of the initial instrument based on the difference evaluation values ​​of all core indicators.

[0106] Step 403: Determine the proportion of core indicators among all instrument indicators.

[0107] Step 404: Calculate the functional indicator evaluation value based on the average value and quantity percentage of the core indicators.

[0108] First, for each core performance indicator of the initial instrument, the manufacturer's claimed value in the instrument's performance indicator data is compared with the benchmark value in the instrument's performance indicator data, and a difference assessment value is calculated according to the indicator type. The purpose of this step is to quantify the degree of deviation between the instrument's claimed performance and the standard requirements.

[0109] Specifically, the calculation method for the difference assessment value differs for different types of indicators.

[0110] For indicators that are considered "below average" (i.e., indicators where smaller values ​​are better), the difference assessment value... ,in The manufacturer's claimed value. Let be the baseline value, i represent the i-th initial instrument, and j represent the j-th core indicator. Where, when At that time, then order ;when At that time, then order .

[0111] For indicators that are considered superior (i.e., indicators where a higher value equates to better performance), the difference assessment value... Among them, when At that time, then order ;when At that time, then order .

[0112] For a specified value-type indicator, the difference assessment value Among them, when At that time, then order ;when At that time, then order .

[0113] For range-specific indicators, i.e., indicators whose values ​​must fall within a specific range, when the claimed value... Falling within the benchmark range Internal time, difference assessment value A value of 1 indicates a perfect match; where, As the lower limit of the benchmark range, This is the upper limit of the baseline range. Otherwise, the difference assessment value is calculated based on the degree of deviation from the range: when hour, ,when hour, And when hour, ;when hour, .

[0114] Based on this, the evaluation values ​​are determined according to the differences in all core indicators. Calculate the average value of the core performance indicators of the initial instrument. The mean of the core indicator evaluation reflects the overall performance level of the initial instrument after covering the core indicators. The mean of the core indicator evaluation ranges from (0,1), where a mean value closer to 1 indicates better overall performance. The specific calculation of the mean of the core indicator evaluation is as follows: The formula is as follows:

[0115]

[0116] in, This represents the number of core indicators for the i-th initial instrument, which is the total number of indicators in the standard core indicator set actually included in the i-th initial instrument. This is the difference assessment value of the j-th core indicator for the i-th initial instrument.

[0117] Then, to avoid overestimating the performance of instruments that only cover a few core metrics but perform exceptionally well, it is necessary to measure the completeness of the initial instrument's coverage of the standard core metric set, that is, to determine the proportion of core metrics in all instrument metrics. Percentage of Quantities The value range is (0,1), where the closer the percentage is to 1, the more comprehensive the coverage.

[0118] Specifically, determining the proportion of quantities The specific methods are as follows:

[0119] First, determine the set of standard core indicators. Based on industry standards or application scenario requirements, such as the "Performance Test Methods for Atomic Absorption Spectrophotometers," define the total library of indicators required for similar instruments with the same function, i.e., all instrument indicators, denoted as N. For example, the standard core indicator set for atomic absorption spectrophotometers used for water quality testing includes four instrument indicators: detection limit, precision, accuracy, and linear range, so N=4.

[0120] Next, we will count the number of core indicators for the i-th initial instrument. Verify the technical parameters of the i-th initial instrument and count the number of core indicators it actually includes. For example, if the core indicators of the i-th initial instrument are the limit of detection, precision, and linear range, then... =3.

[0121] Then according to the formula Calculate the percentage of indicators Based on the examples above, That is, the i-th initial instrument on the surface covers 75% of the core indicators.

[0122] Finally, the average value of the core indicators will be evaluated. and the proportion of quantity Multiply to obtain the functional indicator evaluation value. ,Right now, The calculated functional index evaluation values ​​can simultaneously take into account the performance of existing indicators and the completeness of core indicator coverage, and can comprehensively reflect the overall functional performance level of the initial instrument. The value range of the functional index evaluation value is (0,1). The closer the functional index evaluation value is to 1, the better the function of the initial instrument.

[0123] Step 302: Calculate the price assessment value of each initial instrument based on the manufacturer's instrument data and the winning bid data.

[0124] Specifically, this application further calculates the initial instrument price assessment value from the perspective of instrument price.

[0125] The price of an instrument is also a key factor in the selection decision. However, in practical applications, manufacturers' quoted prices often differ from actual market transaction prices, and market prices themselves fluctuate. Therefore, how to assess whether an instrument's quoted price is reasonable and quantify the gap between it and its true market value is a technical problem that urgently needs to be solved in the procurement and selection process of scientific instruments based on artificial intelligence.

[0126] In step 302, the manufacturer's instrument data includes the instrument price information displayed by the manufacturer, such as the price range. The winning bid data comes from public platforms such as government procurement websites and records the transaction prices of the instruments in actual procurement projects. The price assessment value is a numerical value used to quantify the difference between the instrument's quoted price and the actual market transaction price.

[0127] In one specific implementation, firstly, genuine transaction records related to the target instrument are filtered from a massive amount of winning bid data, and based on these, the average market transaction price of the instrument within a certain time range is calculated. Then, the manufacturer's quoted price is obtained. Finally, by comparing the manufacturer's quoted price with the calculated average market transaction price, an assessment value reflecting the reasonableness of its price is obtained.

[0128] In a preferred embodiment, refer to Figure 5 , Figure 5 This is the fifth flowchart of the artificial intelligence-based scientific instrument procurement and selection method provided by the present invention. The specific process of calculating the price assessment value of each initial instrument includes steps 501 to 507:

[0129] Step 501: Filter the valid winning bid records from the winning bid data that match the initial instrument and are within the preset time range, and remove outliers that deviate from the average price of similar instruments by more than a preset standard deviation multiple.

[0130] Step 502: Calculate the average price of valid winning bids as the average market winning bid price for the initial instrument.

[0131] Step 503: Extract the initial instrument manufacturer's displayed price range from the manufacturer's instrument data.

[0132] Step 504: Calculate the manufacturer's quoted price based on the price range displayed by the manufacturer.

[0133] Step 505: Obtain the bid difference value based on the difference between the manufacturer's quoted price and the average market bid price.

[0134] Step 506: Determine the initial instrument's bid fluctuation value based on the difference between the maximum and minimum values ​​among all valid bid records.

[0135] Step 507: Obtain the price assessment value based on the relationship between the difference in the winning bid and the fluctuation value of the winning bid.

[0136] To obtain representative market transaction price data, firstly, valid winning bid records that match the initial instrument and are within a preset time range are selected from the winning bid data, and outliers that deviate from the average price of similar instruments by more than a preset standard deviation are removed.

[0137] Specifically, firstly, based on unique identifiers such as instrument brand or model, all relevant winning bid records are located from a massive amount of winning bid data. Then, a preset time range is set, such as the past three years, to ensure data timeliness. Finally, to eliminate interference from related-party transactions or abnormal pricing, outlier removal is necessary. One feasible removal method is to calculate the average price and standard deviation σ of all winning bid prices for the same type of instrument, and to consider records deviating from the average price by more than a preset standard deviation multiple (e.g., ±3σ) as outliers and remove them; the remaining records are the valid winning bid records.

[0138] After obtaining valid bid records, the average price of these valid bid records is calculated, and this average price is used as the initial market average bid price for the instrument. The market average bid price is a core indicator reflecting the instrument's value in the actual procurement market. Its calculation formula is as follows: ,in, The average market winning bid price for instrument i is represented by t, where t is the total number of valid winning bids. This represents the winning bid price for the kth valid winning bid record.

[0139] At the same time, the manufacturer's displayed price range for the initial instrument is extracted from the manufacturer's instrument data. This price range is a price range displayed by the manufacturer on public channels for potential customers to reference, and usually includes a minimum price and a maximum price.

[0140] Next, based on the price range extracted from the manufacturers' displays, calculate the manufacturers' quoted prices. This step is to obtain a single value that represents the manufacturers' pricing level for subsequent comparison. A simple calculation method is to take the average of their displayed price ranges, and the formula is as follows: ,in, This is the calculated reference value for the manufacturer's external quotation.

[0141] Furthermore, the difference between the manufacturer's quoted price and the average market bid price is used to obtain the bid difference value. This bid difference value directly quantifies the gap between the manufacturer's quoted price and the actual market value. Its calculation formula is as follows: The difference obtained This is the bid difference value for the instrument. If the value is positive, it means that the manufacturer's quoted price is higher than the average market price; if it is negative, it means that it is lower than the average market price; if it is close to zero, it means that the quoted price is basically in line with the market price.

[0142] Next, based on the maximum value among all valid winning bid records and minimum value The difference determines the initial instrument's bid fluctuation value. .Right now, .

[0143] Winning bid floating value Used to reflect the range of market price fluctuations, as a benchmark range for normalization, it can eliminate the impact of the total price difference of different instruments on the valuation. For example, a price difference of 1,000 yuan has different weights in instruments priced at 100,000 yuan and instruments priced at 1 million yuan.

[0144] Finally, based on the difference in bid winnings... With the winning bid floating value The size relationship is used to obtain the price assessment value. .

[0145] Specifically, when the difference between the bids is less than or equal to a negative bid fluctuation value, that is... At this time, the price competitiveness is strongest, making ;

[0146] When the difference between the bids is greater than or equal to the floating value of the bid, that is... At this time, price competitiveness is at its weakest, making ;

[0147] When the bid difference falls between a negative and a positive bid fluctuation value, the price assessment value is calculated using a linear ratio. Among them, the denominator It is the complete fluctuation range from the lowest price to the highest price; numerator This represents the distance between the manufacturer's quoted price and the lowest price. The closer the quoted price is to or below the average price, the lower it is considered to be. The closer it is to 1.

[0148] It should be noted that for cases with no valid winning bid records, i.e., when t=0; for example, if the brand or model of the instrument corresponding to the i-th initial instrument has no winning bid records in the past 3 years, the average winning bid price of "instruments with the same function and accuracy level" can be referenced, or the price can be directly... Set the value to 0.5 for neutral evaluation and note in the results that "there is no directly winning data, and the evaluation value is a reference value".

[0149] For the case where all winning bid floating values ​​are the same, i.e. For example, if all valid winning bids have the same winning bid price, that is... ,but At this time, if Then let ;like Then let .

[0150] Step 303: Calculate the user evaluation value for each initial instrument based on user discussion data and user relationship data.

[0151] Specifically, this application ultimately calculates the initial user evaluation value of the instrument from the perspective of user evaluation.

[0152] The performance of an instrument in actual use and user reviews are also key factors in the selection decision. However, in practice, the quality of online user discussions varies greatly, with many unprofessional, subjective, or even false reviews. Directly adopting this information may mislead the selection decision. Therefore, how to extract objective and reliable comprehensive evaluations from massive amounts of user discussions is a pressing technical problem that needs to be solved in the procurement and selection process of scientific instruments based on artificial intelligence.

[0153] In step 303, the user discussion data includes user reviews of a specific instrument. The user relationship data records user interaction behavior. The user evaluation value is a numerical value used to quantify the overall reputation of an instrument among a real user group.

[0154] In one specific implementation, user relationship data is first analyzed to assess the relative credibility or influence of each user who posts a comment. Then, the content of comments in user discussion data is analyzed to determine whether the sentiment expressed is positive, negative, or neutral. Finally, the sentiment of the comment is combined with the credibility of its poster for a comprehensive calculation, resulting in an evaluation value that represents the overall user reputation of the instrument. During the aggregate calculation, comments from users with higher credibility are given higher weight.

[0155] In a preferred embodiment, refer to Figure 6 , Figure 6 This is the sixth flowchart of the scientific instrument procurement and selection method based on artificial intelligence provided by the present invention. The specific process of calculating the user evaluation value of each initial instrument includes steps 601 to 605.

[0156] Step 601: Construct a user attention network based on user relationship data, where users are nodes in the user attention network and the attention relationships between users are directed edges in the user attention network.

[0157] Step 602: Iteratively calculate the credibility value for each user based on the network the user follows.

[0158] Step 603: Perform sentiment analysis on each comment in the user discussion data to obtain a sentiment score.

[0159] Step 604: For any comment, multiply the sentiment score by the credibility value of the user who posted the comment to obtain a weighted score.

[0160] Step 605: For any initial instrument, calculate the average of the weighted scores of all comments to obtain the user evaluation value.

[0161] First, a user follow network is constructed based on user relationship data, where users are nodes and the follow relationships between users are directed edges. The purpose of this step is to model the abstract user social relationships, laying the foundation for subsequent evaluations of user influence or credibility. Specifically, each user in the database is mapped to a node in the network. If user A follows user B, a directed edge is created in the network pointing from node A to node B. In this way, a user followed by many users will have a large number of incoming edges in the network.

[0162] Subsequently, based on the aforementioned user attention network, the credibility value of each user is iteratively calculated. This step employs the PageRank algorithm, treating a user's credibility as their importance within the user attention network. A user's credibility primarily derives from the credibility passed down by other users who follow them. The iterative calculation formula is as follows:

[0163]

[0164] in, This represents the credibility value of user u at the (t+1)th iteration; d represents the credibility value of user v in the t-th iteration; d is the damping coefficient, usually taken as 0.85, representing the probability that the user randomly browses and follows other users; M is the total number of users in the network. It is the set of all users who follow user u; This represents the total number of people followed by user v. All users initially have equal credibility. This iterative process continues until the change in the credibility value of all users in two consecutive iterations is less than a very small preset threshold, for example... At this point, the results are considered to have converged, and the final credibility value for each user is obtained.

[0165] After obtaining user credibility scores, sentiment analysis is performed on each comment in the user discussion data to obtain a sentiment score. The purpose of this step is to quantify the sentiment polarity expressed by a single comment. Specifically, natural language processing techniques are used to analyze the comment text and categorize it as positive, neutral, or negative. For example, comments expressing positive experiences, such as "The instrument is highly accurate and very easy to use," are assigned a sentiment score. For neutral comments, such as "the instrument's appearance design is average," a sentiment score is assigned. For comments expressing negative experiences, such as "multiple malfunctions, poor experience", an emotion score is assigned. .

[0166] Then, for any comment, the aforementioned sentiment score is multiplied by the credibility score of the user who posted the comment to obtain a weighted score. This step links a user's statement to their own credibility, giving higher-credibility users' evaluations a greater weight in the final result. The calculation formula is as follows: ,in, The weighted score of a single comment by user u on instrument i; The credibility value of user u calculated in the preceding steps is, i.e. ; Assign an emotional score to this comment.

[0167] Finally, for any initial instrument, the weighted average score of all relevant comments is calculated to obtain the user evaluation score for that instrument. This step aggregates all credibility-weighted comment scores for that instrument to form the final comprehensive user reputation score. Before calculation, comments can be preprocessed to filter out spam comments lacking substance. The calculation formula is as follows: = ,in, Let be the overall user evaluation value of the i-th initial instrument, ranging from -1 to 1. A value closer to 1 indicates a more positive overall user evaluation, while a value closer to -1 indicates a more negative overall user evaluation. n represents the total number of valid reviews obtained for that instrument. Finally, the overall user evaluation values ​​of all initial instruments are normalized to obtain the user usage evaluation value for each initial instrument, mapping each user usage evaluation value to a specified interval [0, 1], thus facilitating subsequent weighted summation. The specific normalization method is not specified in this application.

[0168] Step 103: Match the selection requirements information input by the user in the knowledge graph to obtain a set of candidate instruments.

[0169] Because users often express their needs in colloquial, unstructured natural language, such as "I want to find an imported atomic absorption spectrometer to measure heavy metals in food," traditional keyword-based search methods struggle to accurately understand the deep semantics of such complex requests, easily leading to inaccurate matching results or missing information. To solve the technical challenge of translating vague user needs into precise instrument matching, a method is needed that can deeply understand user intent and accurately match candidate instruments from the massive data of knowledge graphs.

[0170] The selection requirements information consists of textual information submitted by the user through the input area, describing their purchasing intentions. The knowledge graph is a structured knowledge network constructed in the preceding steps, containing various entities and their relationships. The candidate instrument set refers to one or more sets of instruments that, after initial screening in the knowledge graph, meet the user's selection requirements information.

[0171] Specifically, the selection requirements information in natural language input by the user is first semantically parsed to extract key information points, such as application industry, testing items, instrument principles, and desired brand or place of origin.

[0172] Subsequently, this extracted key information is used as query criteria for retrieval within the constructed knowledge graph. This retrieval process involves not only simple text matching but may also involve semantic similarity calculations.

[0173] Finally, the instrument entities with high matching degrees and their associated information in the knowledge graph are filtered out to form a preliminary set of candidate instruments.

[0174] In one specific implementation, refer to Figure 7 , Figure 7 This is the seventh flowchart of the artificial intelligence-based scientific instrument procurement and selection method provided by the present invention. Step 103 specifically includes steps 701 to 703:

[0175] Step 701: Parse the selection requirement information through a large language model to extract at least one key piece of information, and generate corresponding key-value pairs based on the key information.

[0176] Step 702: Using key-value pairs as query conditions, perform a retrieval in the knowledge graph. For entities containing vector value attributes, calculate the similarity between the vector of the corresponding field in the query conditions and the entity attribute vector in the knowledge graph.

[0177] Step 703: Filter out entities with similarity greater than a preset threshold, and sort them according to the association rules between entities and the user attention rules to form a candidate instrument set.

[0178] Specifically, firstly, the selection requirements information is analyzed using a large language model to extract at least one key piece of information, and corresponding key-value pairs are generated based on this key information. When a user submits a natural language description in the input area, such as "second-hand imported ICP-MS for measuring heavy metals such as lead, cadmium, and chromium in grains, for use in Chongqing," directly using keyword search cannot fully and accurately capture their complete intent. This step utilizes the natural language understanding capabilities of the large language model to solve this problem.

[0179] Specifically, the text information entered by the user is sent to a designated large language model service interface. The model analyzes the text according to the preset prompt word rules, identifies and extracts the structured information contained therein, such as application industry, principle method, test sample, test item, place of origin, manufacturer, price, brand, model, etc.

[0180] Then, this extracted information is assigned to predefined keys, forming key-value pairs. For example, for the input above, key-value pairs such as {"application_industry":"grain and oil","test_item":["chromium","heavy metals","pesticide residues"],"instrument_principle":"ICP-MS","origin_place":"imports","usage_location":"Chongqing"} might be generated.

[0181] After generating key-value pairs, these key-value pairs are used as query criteria to perform retrieval in the knowledge graph.

[0182] This step applies the structured query criteria generated in the previous step to the knowledge graph to find relevant entities. This retrieval process is a hybrid query. For attributes that can be directly matched (such as place of origin, brand, etc.), precise attribute value matching is performed. For attributes containing semantic information, similarity calculations are performed using vectors.

[0183] Specifically, for entities containing vector-valued attributes, it is necessary to calculate the similarity between the vector of the corresponding field in the query conditions and the entity attribute vector in the knowledge graph. For example, for the query condition "test_item":"chromium content", first obtain the query vector corresponding to "chromium content", and then perform similarity calculation (e.g., cosine similarity calculation) between this vector and the "application scenario vector value" or "indicator item vector value" of the "standard instrument entity" in the knowledge graph, thereby finding the semantically closest entity.

[0184] Finally, entities with similarity greater than a preset threshold are filtered out and sorted according to entity association rules and user attention rules to form a candidate instrument set. The purpose of this step is to refine and sort the preliminary search results to improve the accuracy and relevance of the final results.

[0185] Specifically, a similarity threshold is first set, such as 0.8. Only entities with a similarity score greater than this threshold will be retained.

[0186] Next, the results are expanded using the association rules between entities. For example, if a "standard instrument entity" is matched, its associated "manufacturer instrument entity" and "standard entity" are automatically included in the result set based on the association relationships in the knowledge graph.

[0187] Finally, the result set can be sorted according to user attention rules, for example, placing entities with higher user viewing priority at the top. After filtering, expansion, and sorting, the final list of instruments is the candidate instrument set.

[0188] Step 104: Extract the evaluation values ​​of each candidate instrument in the candidate instrument set across all evaluation dimensions.

[0189] Specifically, for each instrument in the candidate instrument set, its evaluation values ​​across all evaluation dimensions are retrieved from its corresponding manufacturer's instrument entity or associated data table. For example, the instrument's functional performance evaluation values, price evaluation values, and user usage evaluation values ​​are retrieved. Then, these extracted evaluation values ​​are associated with the instrument's basic information to form a dataset containing multi-dimensional quantitative scores for use in subsequent steps.

[0190] Step 105: Calculate the weighted average of the evaluation values ​​of all candidate instruments based on the dimension weight parameters set by the user to obtain a personalized comprehensive score.

[0191] Because different users often prioritize different factors when making purchasing decisions—for example, research institutions may place greater emphasis on the functional performance of instruments, while third-party testing agencies may be more price-sensitive—a fixed, universally applicable recommendation ranking cannot meet such diverse selection needs. To address the mismatch between recommended results and users' actual preferences, a mechanism is needed to allow the selection method to dynamically adjust based on users' personalized needs. Therefore, after extracting the evaluation values ​​of each candidate instrument, this invention also needs to perform a weighted calculation on the evaluation values ​​of all candidate instruments according to the dimension weight parameters set by the user, obtaining a personalized comprehensive score.

[0192] The dimension weight parameter is a set of values ​​used to represent the degree of importance that users attach to different evaluation dimensions. This dimension weight parameter is set by the user according to their own procurement needs. For example, users can set weights for the functional indicator evaluation dimension, the price evaluation dimension, and the user usage evaluation dimension, and the sum of all weights is usually set to 1 or 100%.

[0193] The personalized composite score is a final value that takes into account user preferences and is used to comprehensively measure the overall quality of candidate instruments.

[0194] In one specific implementation, the weight parameters set by the user for each evaluation dimension are first obtained through a user interface (such as a slider or input box).

[0195] Then, for each candidate instrument in the candidate instrument set, its evaluation value under each evaluation dimension is multiplied by its corresponding dimension weight parameter, and all products are added together. The sum is the personalized comprehensive score of the candidate instrument.

[0196] Step 106: Based on the personalized comprehensive score and the association information in the knowledge graph, generate a selection scheme that includes instrument recommendation results and data traceability information.

[0197] Specifically, the instrument recommendation results are usually a list of candidate instruments sorted from high to low based on the personalized comprehensive score.

[0198] Related information refers to various types of information that are directly or indirectly connected to the candidate instrument entity in the knowledge graph, such as the specific standard documents followed by the instrument, details of historical bidding records, and related user comments.

[0199] Data traceability information refers to information that establishes a clear correspondence between the recommendation results and the original data source, such as providing web links to specific winning bid announcements or download addresses for attachments to standard documents.

[0200] The selection plan is the final structured report presented to the user, which combines quantitative ranking, qualitative analysis, and traceable evidence.

[0201] In one specific implementation, the candidate instruments are first sorted according to their individual comprehensive scores to form a recommendation list.

[0202] Then, for each instrument in the recommendation list, the association information related to its evaluation value calculation is extracted by traversing the edges and nodes connected to the instrument entity in the knowledge graph. For example, the specific winning bid record number and link used to calculate the price evaluation value are extracted, the source standard number used to calculate the functional indicator evaluation value is extracted, and the specific user review content used to calculate the user usage evaluation value is extracted.

[0203] Finally, following the preset report template, the recommended ranking of instruments, evaluation values ​​of each dimension, personalized comprehensive scores, and all extracted data traceability information are integrated. The large language model can be called to analyze this structured data, generate qualitative comparison descriptions and selection suggestions, and finally form a complete selection plan.

[0204] It should be noted that this plan will also update the evaluation values ​​of each instrument in real time based on the relevant data added or modified for each instrument, combined with real-time market prices and the latest user feedback, so that the final selection plan can be continuously adjusted according to actual data, thereby achieving the accuracy of the recommended selection plan.

[0205] Reference Figure 8 , Figure 8 This is a schematic diagram of the structure of the artificial intelligence-based scientific instrument procurement and selection device provided by the present invention. The device includes:

[0206] The first processing module is used to construct a knowledge graph based on the relevant data from all initial instruments under various data sources;

[0207] The second processing module is used to calculate the evaluation value of each initial instrument under each evaluation dimension based on at least one of the relevant data and multiple preset evaluation dimensions.

[0208] The third processing module is used to match the selection requirements information input by the user in the knowledge graph to obtain a set of candidate instruments;

[0209] The fourth processing module is used to extract the evaluation values ​​of each candidate instrument in the candidate instrument set under all the evaluation dimensions;

[0210] The fifth processing module is used to perform weighted calculations on the evaluation values ​​corresponding to all the candidate instruments according to the dimension weight parameters set by the user, so as to obtain a personalized comprehensive score;

[0211] The sixth processing module is used to generate a selection scheme that includes instrument recommendation results and data traceability information based on the personalized comprehensive score and the association information in the knowledge graph.

[0212] It should be noted that the artificial intelligence-based scientific instrument procurement and selection device provided by the present invention can execute the artificial intelligence-based scientific instrument procurement and selection method of any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0213] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 9 As shown, the electronic device may include: a processor 910, a communication interface 920, a memory 930, and a communication bus 940. The processor 910, communication interface 920, and memory 930 communicate with each other via the communication bus 940. The processor 910 can call logical instructions in the memory 930 to execute an artificial intelligence-based scientific instrument procurement and selection method. This method includes: constructing a knowledge graph based on relevant data from various data sources for all initial instruments; calculating the evaluation value of each initial instrument in each evaluation dimension based on at least one relevant data source and multiple preset evaluation dimensions; matching user-input selection requirements in the knowledge graph to obtain a set of candidate instruments; extracting the evaluation values ​​of each candidate instrument in the candidate instrument set across all evaluation dimensions; weighting the evaluation values ​​of all candidate instruments according to user-set dimension weight parameters to obtain a personalized comprehensive score; and generating a selection scheme containing instrument recommendation results and data traceability information based on the personalized comprehensive score and the association information in the knowledge graph.

[0214] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0215] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer is able to execute the artificial intelligence-based scientific instrument procurement and selection method provided in the above embodiments.

[0216] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the artificial intelligence-based scientific instrument procurement and selection method provided in the above embodiments.

[0217] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0218] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0219] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A scientific instrument procurement and selection method based on artificial intelligence, characterized in that, include: A knowledge graph is constructed based on the relevant data from all initial instruments across various data sources. Based on at least one of the aforementioned relevant data and multiple preset evaluation dimensions, an evaluation value for each initial instrument is calculated under each evaluation dimension; the evaluation dimensions include functional indicator dimensions; wherein, the relevant data includes manufacturer instrument indicator data and instrument measurement indicator data; the calculation of the evaluation value for each initial instrument under each evaluation dimension based on at least one of the aforementioned relevant data and multiple preset evaluation dimensions includes: Based on the manufacturer's instrument performance data and the instrument measurement data, calculate the functional performance evaluation value for each initial instrument; For each core indicator of the initial instrument, the manufacturer's claimed value in the manufacturer's instrument indicator data is compared with the benchmark value in the instrument measurement indicator data, and the difference assessment value is calculated according to the indicator type. Calculate the mean value of the core indicators of the initial instrument based on the difference evaluation values ​​of all the core indicators; Determine the percentage of the core indicators among all instrument indicators; The functional indicator evaluation value is calculated by multiplying the average value of the core indicator evaluation and the percentage of the quantity. The selection requirements input by the user are matched in the knowledge graph to obtain a set of candidate instruments; Extract the evaluation value of each candidate instrument in the candidate instrument set under all the evaluation dimensions; The evaluation values ​​of all candidate instruments are weighted according to the dimension weight parameters set by the user to obtain a personalized comprehensive score. Based on the personalized comprehensive score and the association information in the knowledge graph, a selection scheme is generated that includes instrument recommendation results and data traceability information.

2. The method for purchasing and selecting scientific instruments based on artificial intelligence according to claim 1, characterized in that, The relevant data includes standard information data, instrument measurement index data, manufacturer instrument data, manufacturer instrument index data, winning bid data, user discussion data, and user relationship data; The construction of a knowledge graph based on relevant data from all initial instruments across various data sources includes: Based on the standard information data, the instrument measurement index data, the manufacturer instrument data, the manufacturer instrument index data, the winning bid data, the user discussion data, and the user relationship data, define the standard entity, the standard instrument entity, and the manufacturer instrument entity; A first association is established between the standard instrument entity and the manufacturer instrument entity through instrument information and indicator names; a second association is established between the manufacturer instrument entity and the manufacturer instrument indicator data through instrument information; a third association is established between the manufacturer instrument entity and the winning bid data through instrument information; and a fourth association is established between user discussion data and user relationship data through user ID. The knowledge graph is constructed based on the standard entity, the standard instrument entity, the manufacturer instrument entity, the first association, the second association, the third association, and the fourth association.

3. The method for purchasing and selecting scientific instruments based on artificial intelligence according to claim 2, characterized in that, The standard entity contains all fields of the standard information data, as well as standard name vector values ​​and industry vector values; the standard instrument entity contains all fields of the instrument measurement index data, as well as instrument name vector values, application industry vector values, index item vector values, principle method vector values, and application scenario vector values; the manufacturer instrument entity contains all fields of the manufacturer instrument data, as well as instrument name vector values, brand vector values, manufacturer name vector values, and place of origin vector values.

4. The method for purchasing and selecting scientific instruments based on artificial intelligence according to claim 2, characterized in that, Based on at least one of the aforementioned relevant data and multiple preset evaluation dimensions, calculate the evaluation value of each initial instrument under each evaluation dimension, including: Based on the manufacturer's instrument performance data and the instrument measurement data, calculate the functional performance evaluation value for each initial instrument; Based on the manufacturer's instrument data and the winning bid data, calculate the price assessment value for each initial instrument; Based on the user discussion data and the user relationship data, calculate the user usage evaluation value for each initial instrument.

5. The method for purchasing and selecting scientific instruments based on artificial intelligence according to claim 4, characterized in that, The step of calculating the price assessment value of each initial instrument based on the manufacturer's instrument data and the winning bid data includes: Valid winning bid records that match the initial instrument and are within a preset time range are selected from the winning bid data, and outliers that deviate from the average price of similar instruments by more than a preset standard deviation multiple are removed. The average price of valid winning bids is calculated as the market average winning bid price for the initial instrument. Extract the manufacturer's displayed price range for the initial instrument from the manufacturer's instrument data; Calculate the manufacturer's quoted price based on the price range displayed by the manufacturer; The difference between the manufacturer's quoted price and the average winning bid price in the market is used to obtain the winning bid difference value; The initial instrument's bid fluctuation value is determined based on the difference between the maximum and minimum values ​​among all the valid bid-winning records. The price assessment value is obtained based on the relationship between the difference in the winning bid and the fluctuation value of the winning bid.

6. The method for purchasing and selecting scientific instruments based on artificial intelligence according to claim 4, characterized in that, The step of calculating the user usage evaluation value for each initial instrument based on the user discussion data and the user relationship data includes: A user attention network is constructed based on user relationship data, where users are nodes in the user attention network and the attention relationships between users are directed edges in the user attention network; Based on the user attention network, the credibility value of each user is iteratively calculated; Sentiment analysis is performed on each comment in the user discussion data to obtain a sentiment score; For any of the aforementioned comment contents, the sentiment score is multiplied by the credibility value of the user who published the comment content to obtain a weighted score; For any of the initial instruments, the weighted average of the scores of all the comments is calculated to obtain the user usage evaluation value.

7. The method for purchasing and selecting scientific instruments based on artificial intelligence according to claim 1, characterized in that, The process involves matching the user's input selection requirements within the knowledge graph to obtain a set of candidate instruments, including: The selection requirements information is parsed using a large language model to extract at least one key piece of information, and corresponding key-value pairs are generated based on the key information. Using the key-value pairs as query conditions, a retrieval is performed in the knowledge graph, wherein, for entities containing vector value attributes, the similarity between the vector of the corresponding field in the query conditions and the entity attribute vector in the knowledge graph is calculated; Entities with similarity greater than a preset threshold are filtered out and sorted according to the association rules between entities and the user attention rules to form the candidate instrument set.

8. A scientific instrument procurement and selection device based on artificial intelligence, characterized in that, include: The first processing module is used to construct a knowledge graph based on the relevant data from all initial instruments under various data sources; The second processing module is used to calculate the evaluation value of each initial instrument under each evaluation dimension based on at least one of the aforementioned relevant data and multiple preset evaluation dimensions; the evaluation dimensions include functional indicator dimensions; wherein, the relevant data includes manufacturer instrument indicator data and instrument measurement indicator data; calculating the evaluation value of each initial instrument under each evaluation dimension based on at least one of the aforementioned relevant data and multiple preset evaluation dimensions includes: calculating the functional indicator evaluation value of each initial instrument based on the manufacturer instrument indicator data and the instrument measurement indicator data; for each core indicator of the initial instrument, comparing the manufacturer's claimed value in the manufacturer instrument indicator data with the benchmark value in the instrument measurement indicator data, and calculating the difference evaluation value according to the indicator type; calculating the core indicator evaluation mean of the initial instrument based on the difference evaluation values ​​of all the core indicators; determining the proportion of the core indicator in all instrument indicators; and calculating the functional indicator evaluation value based on the product of the core indicator evaluation mean and the proportion. The third processing module is used to match the selection requirements information input by the user in the knowledge graph to obtain a set of candidate instruments; The fourth processing module is used to extract the evaluation values ​​of each candidate instrument in the candidate instrument set under all the evaluation dimensions; The fifth processing module is used to perform weighted calculations on the evaluation values ​​corresponding to all the candidate instruments according to the dimension weight parameters set by the user, so as to obtain a personalized comprehensive score; The sixth processing module is used to generate a selection scheme that includes instrument recommendation results and data traceability information based on the personalized comprehensive score and the association information in the knowledge graph.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the artificial intelligence-based scientific instrument procurement and selection method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the artificial intelligence-based scientific instrument procurement and selection method as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the artificial intelligence-based scientific instrument procurement and selection method as described in any one of claims 1 to 7.

Images