Method and system for constructing digital dynamic map to describe the diffusion of literature influence

Abstract

The present invention belongs to the field of artificial intelligence technology, and relates to a method and system for constructing a digital dynamic map that describes the diffusion of the influence of documents. The method includes: using an RPA-based robot to construct a document automated query robot; constructing a document automated reading agent robot based on a large model; calculating a first correlation coefficient; constructing and running an image automated comparison agent robot based on a large model, and calculating a second correlation coefficient; constructing and running a document keyword automated excerpting agent robot based on a large model; constructing a digital map automated annotation robot to obtain a digital dynamic map that describes the diffusion of the influence and development trend of the solutions contained in known documents. The present invention proposes to realize intelligent and timely tracking of the dissemination and diffusion of the solutions contained in the documents in an unattended mode based on RPA combined with an agent robot based on a large model, which has high academic value, commercial value and strong scalability.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular, relates to a method and system for constructing a digital dynamic map that describes the diffusion of document influence. Background Art

[0002] In the existing technology, with the help of Stanford AI Town open source simulated virtual hospital, patients, nurses and doctors are all intelligent agents driven by large language models, simulating the key links of the entire closed cycle of treating patients' diseases in the virtual hospital: onset, triage, registration, consultation, physical examination, diagnosis, medication, rehabilitation and post-hospital follow-up. Intelligent doctors can iterate and evolve with only a small amount of real case data without manually labeled data, continuously improving treatment performance over time and finally achieving high diagnostic accuracy.

[0003] With the growing development of artificial intelligence, manually screening, reading and annotating documents only consumes huge human resources and makes it difficult to achieve real-time tracking of documents. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method and system for constructing a digital dynamic map that describes the diffusion of document influence.

[0005] In a first aspect, the present invention provides a method for constructing a digital dynamic map describing the diffusion of document influence, comprising:

[0006] Use RPA robots to build automated document query robots and run automated document query systems. The automated document query robots perform automated document query and download tasks based on keywords.

[0007] Construct an intelligent robot for automated document reading based on a large model, and use the intelligent robot for automated document reading; the automated document reading robot performs document reading tasks, document abstract extraction tasks, and document correlation coefficient calculation tasks;

[0008] Extracting feature vectors of documents in the subject database and known documents respectively, and calculating the first correlation coefficient of the feature vectors of documents in the subject database and known documents;

[0009] Filter the documents in the subject database whose first correlation coefficient with the known documents is greater than a first set value and save them as the first intermediate documents;

[0010] Constructing and running an image automatic comparison agent robot based on a large model to calculate a second correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document;

[0011] Selecting first intermediate documents whose second correlation coefficient with the known document is greater than a second set value from among the first intermediate documents and saving the first intermediate documents as second intermediate documents;

[0012] Construct and run a large-model-based document keyword automatic extraction intelligent agent robot to extract the document keywords in the second intermediate document and save them in the database;

[0013] Use RPA robots to build and run a digital map automatic annotation robot, automatically extract literature keywords from the database and automatically annotate them on the digital map, and obtain a digital dynamic map that describes the influence spread and development trend of the solutions contained in known literature.

[0014] In a second aspect, the present invention provides a digital dynamic map construction system for describing the diffusion of document influence, comprising a query unit, a reading unit, a first correlation coefficient calculation unit, a first screening unit, a second correlation coefficient calculation unit, a second screening unit, an extracting unit and a marking unit;

[0015] The query unit is used to build an automated document query robot using the RPA robot and run the automated document query system. The automated document query robot performs automated document query and download tasks based on keywords.

[0016] The reading unit is used to build an intelligent robot for automatic document reading based on a large model, and use the intelligent robot for automatic document reading; the automatic document reading robot performs document reading tasks, document abstract extraction tasks, and document correlation coefficient calculation tasks;

[0017] A first correlation coefficient calculation unit is used to extract feature vectors of documents in the subject database and known documents respectively, and calculate a first correlation coefficient of the feature vectors of documents in the subject database and known documents;

[0018] A first screening unit is used to screen documents in the subject database and documents whose first correlation coefficient with known documents is greater than a first set value and save them as first intermediate documents;

[0019] A second correlation coefficient calculation unit is used to construct and run an image automatic comparison agent robot based on a large model to calculate a second correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document;

[0020] A second screening unit is used to screen out first intermediate documents whose second correlation coefficient with the known document is greater than a second set value and save them as second intermediate documents;

[0021] An extracting unit is used to construct and run a document keyword automatic extracting intelligent agent robot based on a large model, extract the document keywords in the second intermediate document, and save them in a database;

[0022] The annotation unit is used to use the RPA robot to build and run a digital map automatic annotation robot, automatically extract the literature keywords in the database and automatically annotate them on the digital map, and obtain a digital dynamic map that describes the influence spread and development trend of the solutions contained in the known literature.

[0023] Based on the above technical solution, the present invention can also be improved as follows.

[0024] Furthermore, when the document automatic query robot performs document automatic query and download tasks according to keywords, it searches for documents in the subject database according to the keywords, and downloads documents in the subject database whose titles or abstracts contain the keywords to the subject database.

[0025] Furthermore, the feature vectors of the documents in the subject database and the known documents are extracted respectively, and the cosine similarity of the angle between the feature vectors of the documents in the subject database and the known documents is calculated as the first correlation coefficient; if the cosine similarity of the angle between the feature vectors of the documents in the subject database and the known documents is greater than a first set threshold, the document in the subject database is retained, otherwise the document in the subject database is not retained.

[0026] Furthermore, the second correlation coefficient includes an appearance correlation coefficient and a picture color correlation coefficient; if the appearance correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document is greater than the second set threshold, and the picture color correlation coefficients are greater than the third set threshold, then the first intermediate document is developed from the known document.

[0027] Furthermore, the shape correlation coefficient is calculated according to the scale-invariant feature transformation correlation coefficient algorithm, including:

[0028] Extracting feature points with scale invariance and rotation invariance from the graphics recorded in the first intermediate document and the graphics in the known document respectively; the feature points include position information, scale information and direction information;

[0029] Match the feature points in the two images through the descriptors of the feature points; the descriptors are vectors used to describe the pixel information around the feature points;

[0030] The shape correlation is calculated based on the number and quality of matched feature points.

[0031] Furthermore, a distance measurement method is used to determine the degree of appearance matching between the two images as an appearance correlation coefficient; the distance measurement method includes calculating the Euclidean distance of feature points to determine the appearance similarity of the two images.

[0032] Furthermore, the image color correlation coefficient is calculated according to a color histogram intersection algorithm, including: for each graphic recorded in the first intermediate document and the graphic in the known document, a color histogram is constructed respectively, the frequency of occurrence of each color in the image is counted, and the color histogram intersection coefficient of each graphic recorded in the first intermediate document and the graphic in the known document is calculated as the image color correlation coefficient.

[0033] Furthermore, the literature keywords include the literature abstract keywords, the authors in the literature, the author’s organization, the author’s email address, and the industry to which the literature belongs.

[0034] The beneficial effects of the present invention are as follows: the present invention proposes to combine robot process technology with an intelligent robot based on a large model, integrate visualization technology, and track global documents in an automatic, intelligent and efficient manner, so as to realize intelligent and timely tracking of the global dissemination and diffusion of technical documents in an unattended mode, and has high academic value and commercial value; the present invention has the significant advantages of simple development and deployment, high degree of integration, automated operation, real-time tracking and intelligent analysis, and can track the diffusion and development trend of any new technical document on a global scale by slightly modifying the query keywords, so the present invention has extremely strong scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of a method for constructing a digital dynamic map describing the diffusion of literature influence provided in Example 1 of the present invention;

[0036] Figure 2 This is a schematic diagram of a digital dynamic map construction system for describing the diffusion of document influence provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0038] Example 1

[0039] As an embodiment, in order to solve the above technical problems, as shown in the attached Figure 1 As shown, this embodiment provides a method for constructing a digital dynamic map describing the diffusion of literature influence, including:

[0040] Use RPA (Robotic Process Automation, RPA for short) robots to build automated literature query robots and run automated literature query systems. The automated literature query robots perform automated literature query and download tasks based on keywords.

[0041] Construct an intelligent robot for automated document reading based on a large model, and use the intelligent robot for automated document reading; the automated document reading robot performs document reading tasks, document abstract extraction tasks, and document correlation coefficient calculation tasks;

[0042] Extracting feature vectors of documents in the subject database and known documents respectively, and calculating the first correlation coefficient of the feature vectors of documents in the subject database and known documents;

[0043] Filter the documents in the subject database whose first correlation coefficient with the known documents is greater than a first set value and save them as the first intermediate documents;

[0044] Constructing and running an image automatic comparison agent robot based on a large model to calculate a second correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document;

[0045] Selecting first intermediate documents whose second correlation coefficient with the known document is greater than a second set value from among the first intermediate documents and saving the first intermediate documents as second intermediate documents;

[0046] Construct and run a large-model-based document keyword automatic extraction intelligent agent robot to extract the document keywords in the second intermediate document and save them in the database;

[0047] Use RPA robots to build and run a digital map automatic annotation robot, automatically extract literature keywords from the database and automatically annotate them on the digital map, and obtain a digital dynamic map that describes the influence spread and development trend of known literature.

[0048] Optionally, when the document automatic query robot performs document automatic query and download tasks according to keywords, it searches for documents in the subject database according to the keywords, and downloads documents in the subject database whose titles or abstracts contain the keywords to the subject database.

[0049] In actual application, RPA robots use artificial intelligence-based robotic process automation technology. Robotic process automation technology simulates a series of online operations performed by real people on the Internet. Through a large number of built-in artificial intelligence tools, designers can complete robot operation tasks with simple designs, greatly improving work efficiency and accuracy, and enabling multiple robots to process tasks in parallel.

[0050] By running the automatic document query robot, the automatic document query and download tasks are realized, and the query keywords such as "intelligent body" are used. If the title and abstract of the document contain "intelligent body", the document will be automatically downloaded to the subject database.

[0051] Optionally, feature vectors of documents in the subject database and known documents are extracted respectively, and the cosine similarity of the angle between the feature vectors of documents in the subject database and known documents is calculated as the first correlation coefficient; if the cosine similarity of the angle between the feature vectors of documents in the subject database and known documents is greater than a first set threshold, the document in the subject database is retained, otherwise the document in the subject database is not retained.

[0052] By constructing an intelligent robot for automated document reading based on a large model and using the intelligent robot for automated document reading, we can realize automated execution of document reading tasks, document abstract extraction tasks, and document relevance coefficient calculation tasks for downloaded documents.

[0053] The intelligent robot is built based on a large model and is used to automatically read literature and write literature summaries. It can realize human-like behavior patterns such as perception, decision-making and execution.

[0054] When screening documents in the subject database whose first correlation coefficient with known documents is greater than the first set value and saving them as the first intermediate documents, optionally, taking the Stanford AI Town Open Source New Technology Documents as an example, screen and retain documents in the document database whose correlation coefficient with the Stanford AI Town Open Source New Technology Documents is higher than 60%.

[0055] Optionally, the second correlation coefficient includes an appearance correlation coefficient and a picture color correlation coefficient; if the appearance correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document is greater than the second set threshold, and the picture color correlation coefficients are greater than the third set threshold, then the first intermediate document is developed from the known document.

[0056] In the actual application process, the appearance correlation coefficient and picture color correlation coefficient of the graphics recorded in the retained document and the pictures of the simulation system interface recorded in the Stanford AI Town Open Source New Technology Document are calculated respectively. If both correlation coefficients are higher than 80%, it is determined that the retained document is developed from the "Stanford AI Town Open Source New Technology Document".

[0057] Optionally, the shape correlation coefficient is calculated according to a Scale Invariant Feature Transform (SIFT) correlation coefficient algorithm, including:

[0058] Extracting feature points with scale invariance and rotation invariance from the graphics recorded in the first intermediate document and the graphics in the known document respectively; the feature points include position information, scale information and direction information;

[0059] Match the feature points in the two images through the descriptors of the feature points; the descriptors are vectors used to describe the pixel information around the feature points;

[0060] The shape correlation is calculated based on the number and quality of matched feature points.

[0061] Optionally, a distance measurement method is used to determine the degree of appearance matching between the two images as an appearance correlation coefficient; the distance measurement method includes calculating the Euclidean distance of feature points to determine the appearance similarity of the two images.

[0062] Optionally, the image color correlation coefficient is calculated according to a color histogram intersection algorithm, including: for each graphic recorded in the first intermediate document and the graphic of the known document, a color histogram is constructed respectively, the frequency of occurrence of each color in the image is counted, and the color histogram intersection coefficient of each graphic recorded in the first intermediate document and the graphic of the known document is calculated as the image color correlation coefficient. For example, in the RGB color space, the color range is divided into several intervals, and then the number of pixels in each interval is counted to form a histogram; then the color histogram intersection coefficient of the two images is calculated, and the larger the value of the color histogram intersection coefficient, the more similar the two images are in color distribution.

[0063] Optionally, document keywords include document abstract keywords, authors in the document, author institutions, author email addresses, and the industry to which the document belongs.

[0064] The present invention proposes to combine robot process technology with large-model-based intelligent robots, integrate visualization technology, and track global documents in an automatic, intelligent and efficient manner, so as to realize intelligent and timely tracking of the global dissemination and diffusion of technical documents in an unattended mode, which has high academic and commercial value. The present invention has significant advantages such as simple development and deployment, high degree of integration, automated operation, real-time tracking and intelligent analysis. By slightly modifying the query keywords, the global diffusion and development trend of any new technical document can be tracked, so the present invention has extremely strong scalability.

[0065] Example 2

[0066] Based on the same principle as the method shown in Example 1 of the present invention, as shown in the attached Figure 2As shown, an embodiment of the present invention also provides a digital dynamic map construction system for describing the diffusion of document influence, including a query unit, a reading unit, a first correlation coefficient calculation unit, a first screening unit, a second correlation coefficient calculation unit, a second screening unit, an extracting unit and a marking unit;

[0067] The query unit is used to build an automated document query robot using the RPA robot and run the automated document query system. The automated document query robot performs automated document query and download tasks based on keywords.

[0068] The reading unit is used to build an intelligent robot for automatic document reading based on a large model, and use the intelligent robot for automatic document reading; the automatic document reading robot performs document reading tasks, document abstract extraction tasks, and document correlation coefficient calculation tasks;

[0069] A first correlation coefficient calculation unit is used to extract feature vectors of documents in the subject database and known documents respectively, and calculate a first correlation coefficient of the feature vectors of documents in the subject database and known documents;

[0070] A first screening unit is used to screen documents in the subject database and documents whose first correlation coefficient with known documents is greater than a first set value and save them as first intermediate documents;

[0071] A second correlation coefficient calculation unit is used to construct and run an image automatic comparison agent robot based on a large model to calculate a second correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document;

[0072] A second screening unit is used to screen out first intermediate documents whose second correlation coefficient with the known document is greater than a second set value and save them as second intermediate documents;

[0073] An extracting unit is used to construct and run a document keyword automatic extracting intelligent agent robot based on a large model, extract the document keywords in the second intermediate document, and save them in a database;

[0074] The annotation unit is used to use the RPA robot to build and run a digital map automatic annotation robot, automatically extract the literature keywords in the database and automatically annotate them on the digital map, and obtain a digital dynamic map that describes the influence spread and development trend of known literature.

[0075] Optionally, when the document automatic query robot performs document automatic query and download tasks according to keywords, it searches for documents in the subject database according to the keywords, and downloads documents in the subject database whose titles or abstracts contain the keywords to the subject database.

[0076] Optionally, feature vectors of documents in the subject database and known documents are extracted respectively, and the cosine similarity of the angle between the feature vectors of documents in the subject database and known documents is calculated as the first correlation coefficient; if the cosine similarity of the angle between the feature vectors of documents in the subject database and known documents is greater than a first set threshold, the document in the subject database is retained, otherwise the document in the subject database is not retained.

[0077] Optionally, the second correlation coefficient includes an appearance correlation coefficient and a picture color correlation coefficient; if the appearance correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document is greater than the second set threshold, and the picture color correlation coefficients are greater than the third set threshold, then the first intermediate document is developed from the known document.

[0078] Optionally, the shape correlation coefficient is calculated according to a scale-invariant feature transform correlation coefficient algorithm, including:

[0079] Extracting feature points with scale invariance and rotation invariance from the graphics recorded in the first intermediate document and the graphics in the known document respectively; the feature points include position information, scale information and direction information;

[0080] Match the feature points in the two images through the descriptors of the feature points; the descriptors are vectors used to describe the pixel information around the feature points;

[0081] The shape correlation is calculated based on the number and quality of matched feature points.

[0082] Optionally, a distance measurement method is used to determine the degree of appearance matching between the two images as an appearance correlation coefficient; the distance measurement method includes calculating the Euclidean distance of feature points to determine the appearance similarity of the two images.

[0083] Optionally, the image color correlation coefficient is calculated according to a color histogram intersection algorithm, including: for each graphic recorded in the first intermediate document and the graphic in the known document, respectively constructing a color histogram, counting the frequency of occurrence of each color in the image, and calculating the color histogram intersection coefficient of each graphic recorded in the first intermediate document and the graphic in the known document as the image color correlation coefficient.

[0084] Optionally, document keywords include document abstract keywords, authors in the document, author institutions, author email addresses, and the industry to which the document belongs.

[0085] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for constructing a digital dynamic map describing the diffusion of literature influence, characterized in that: include: Use RPA robots to build automated document query robots and run automated document query systems. The automated document query robots perform automated document query and download tasks based on keywords. Build an intelligent robot for automated document reading based on a large model, and use the intelligent robot for automated document reading; The document automatic reading robot performs document reading tasks, document abstract extraction tasks and document correlation coefficient calculation tasks; Extracting feature vectors of documents in the subject database and known documents respectively, and calculating the first correlation coefficient of the feature vectors of documents in the subject database and known documents; Filter the documents in the subject database whose first correlation coefficient with the known documents is greater than a first set value and save them as the first intermediate documents; Constructing and running an image automatic comparison agent robot based on a large model to calculate a second correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document; Selecting first intermediate documents whose second correlation coefficient with the known document is greater than a second set value from among the first intermediate documents and saving the first intermediate documents as second intermediate documents; Construct and run a large-model-based document keyword automatic extraction intelligent agent robot to extract the document keywords in the second intermediate document and save them in the database; Use RPA robots to build and run a digital map automatic annotation robot, automatically extract literature keywords from the database and automatically annotate them on the digital map, and obtain a digital dynamic map that describes the influence spread and development trend of the solutions contained in known literature.

2. The method for constructing a digital dynamic map describing the diffusion of document influence according to claim 1 is characterized in that: When the document automatic query robot performs document automatic query and download tasks according to keywords, it searches for documents in the subject database according to the keywords, and downloads the documents in the subject database whose titles or abstracts contain the keywords to the subject database.

3. The method for constructing a digital dynamic map describing the diffusion of literature influence according to claim 1 is characterized in that: Extracting feature vectors of documents in the subject database and known documents respectively, and calculating the cosine similarity of the angle between the feature vectors of documents in the subject database and known documents as the first correlation coefficient; If the cosine similarity of the angle between the feature vectors is greater than the first set threshold, the document in the subject database is retained, otherwise the document in the subject database is not retained.

4. The method for constructing a digital dynamic map describing the diffusion of document influence according to claim 1 is characterized in that: The second correlation coefficient includes an appearance correlation coefficient and a picture color correlation coefficient; if the appearance correlation coefficient between the figure recorded in the first intermediate document and the figure contained in the known document is greater than the second set threshold, and the picture color correlation coefficients are greater than the third set threshold, then the first intermediate document is developed from the known document.

5. The method for constructing a digital dynamic map describing the diffusion of document influence according to claim 4 is characterized in that: The shape correlation coefficient is calculated according to the scale-invariant feature transformation correlation coefficient algorithm, including: Extracting feature points with scale invariance and rotation invariance from the graphics recorded in the first intermediate document and the graphics in the known document respectively; the feature points include position information, scale information and direction information; Match the feature points in the two images through the descriptors of the feature points; the descriptors are vectors used to describe the pixel information around the feature points; The shape correlation is calculated based on the number and quality of matched feature points.

6. The method for constructing a digital dynamic map describing the diffusion of document influence according to claim 5 is characterized in that: A distance measurement method is used to determine the degree of appearance matching between two images as an appearance correlation coefficient; the distance measurement method includes calculating the Euclidean distance of feature points to determine the appearance similarity of the two images.

7. The method for constructing a digital dynamic map describing the diffusion of document influence according to claim 4 is characterized in that: The image color correlation coefficient is calculated according to a color histogram intersection algorithm, including: for each graphic recorded in the first intermediate document and the graphic in the known document, a color histogram is constructed respectively, the frequency of occurrence of each color in the image is counted, and the color histogram intersection coefficient of each graphic recorded in the first intermediate document and the graphic in the known document is calculated as the image color correlation coefficient.

8. The method for constructing a digital dynamic map describing the diffusion of document influence according to claim 1 is characterized in that: Literature keywords include literature abstract keywords, authors in the literature, author institutions, author email addresses, and the industry to which the literature belongs.

9. A digital dynamic map construction system describing the diffusion of literature influence, characterized by: It includes a query unit, a reading unit, a first correlation coefficient calculation unit, a first screening unit, a second correlation coefficient calculation unit, a second screening unit, an extracting unit and a marking unit; The query unit is used to build an automated document query robot using the RPA robot and run the automated document query system. The automated document query robot performs automated document query and download tasks based on keywords. The reading unit is used to build an intelligent robot for automatic document reading based on a large model, and use the intelligent robot for automatic document reading; The document automatic reading robot performs document reading tasks, document abstract extraction tasks and document correlation coefficient calculation tasks; A first correlation coefficient calculation unit is used to extract feature vectors of documents in the subject database and known documents respectively, and calculate a first correlation coefficient of the feature vectors of documents in the subject database and known documents; A first screening unit is used to screen documents in the subject database and documents whose first correlation coefficient with known documents is greater than a first set value and save them as first intermediate documents; A second correlation coefficient calculation unit is used to construct and run an image automatic comparison agent robot based on a large model to calculate a second correlation coefficient between the graphics recorded in the first intermediate document and the graphics in the known document; A second screening unit is used to screen out first intermediate documents whose second correlation coefficient with the known document is greater than a second set value and save them as second intermediate documents; An extracting unit is used to construct and run a document keyword automatic extracting intelligent agent robot based on a large model, extract the document keywords in the second intermediate document, and save them in a database; The annotation unit is used to use the RPA robot to build and run a digital map automatic annotation robot, automatically extract the literature keywords in the database and automatically annotate them on the digital map, and obtain a digital dynamic map that describes the influence spread and development trend of known literature.

Images