CN115618947A - Medical knowledge map quality assessment system, device, equipment, medium and product - Google Patents

Abstract

The present disclosure relates to a medical knowledge map quality evaluation system, device, equipment, medium, and product, including: obtaining a medical knowledge map built based on a schema; performing a pre-check operation on the medical knowledge map, and the pre-check operation includes at least Define one of field data type inspection, outlier inspection, and knowledge graph topology inspection; based on social network community discovery and corrected mutual information as an evaluation index, check the medical knowledge graph to obtain the medical knowledge graph. Knowledge graph evaluation results. This application discovers communities existing in different ontologies through pre-checking and the basic algorithm of community discovery in social networks, and uses the revised mutual information to measure the quality of relationships in knowledge graphs.

Description

Medical knowledge map quality assessment system, device, equipment, medium and product

technical field

The present disclosure relates to the field of knowledge graph evaluation, in particular to a medical knowledge graph quality evaluation system, device, equipment, medium and product.

Background technique

At present, there are two main methods for the quality assessment of knowledge graph construction. One is to conduct evaluation by sampling inspection based on manual participation, and the other is a path-based method. The triplet nodes and The relationship is mapped to the vector space, and the vector representation of knowledge is obtained, and the triples in the knowledge graph are evaluated accordingly.

The quality assessment method of knowledge graph based on manual participation not only consumes a lot of human resources but also is very time-consuming, and is greatly affected by the professional background of quality control personnel.

Path-based method There are two main problems with this method: 1. Ignoring the knowledge background, most methods of recording paths treat all relationship types homogeneously, and different relationships and entities are encoded into nodes with only different ids during model input and sides. 2. Most of the large-scale knowledge graphs are built using such methods for embedded representation and completion of knowledge graphs. Using a scheme that is highly similar to the construction method for quality inspection will make it difficult for the model to find interfering data.

Contents of the invention

In order to overcome the problems existing in related technologies, the present disclosure provides a medical knowledge map quality assessment system, device, equipment, medium and product.

According to the first aspect of the embodiments of the present disclosure, a medical knowledge map quality assessment system is provided, including:

Obtain the medical knowledge map built based on the schema;

Performing a pre-check operation on the medical knowledge map, the pre-check operation includes at least one of defined field data type checks, outlier checks, and knowledge map topology checks;

The amount of mutual information based on social network community discovery and correction is used as an evaluation index to check the medical knowledge map, and obtain the evaluation result of the medical knowledge map.

In some embodiments, the definition field data type check includes:

Based on the field type of the medical knowledge map, detect all fields of each entity;

Amend the abnormal result field, and record the proportion of all the abnormal result fields.

In some embodiments, the outlier check includes:

Outlier detection is performed on the point-in degree and point-out degree of each node in the medical knowledge graph, and the outlier points are merged into the graph structure of the medical knowledge graph.

Further, the outlier is merged into the medical knowledge graph structure, including:

The method of keyword matching and edit distance is used to recommend the parent node of the outlier point, and the outlier point is associated with the parent node of the outlier point.

In some embodiments, the knowledge map topology inspection includes:

Directed acyclic graph structure detection using topological sorting;

Use DFS and graph connectivity detection to detect tree structures;

Automatic connection or disconnection processing for abnormal results.

In some embodiments, the amount of mutual information based on social network community discovery and correction is used as an evaluation index to check the medical knowledge map and obtain evaluation results, including:

Based on the basic algorithm in the social network, the communities existing in different ontologies are obtained, and the medical knowledge map is corrected by using the revised mutual information, so as to obtain the evaluation result of the medical knowledge map.

According to the second aspect of the embodiments of the present disclosure, a medical knowledge map quality assessment device is provided, including:

The acquisition module is used to acquire the medical knowledge graph built based on the schema;

A pre-detection module, configured to perform a pre-check operation on the medical knowledge graph, the pre-check operation at least including one of defined field data type checks, outlier checks, and knowledge graph topology checks;

The community network module checks the medical knowledge map based on the social network community discovery and the corrected mutual information as an evaluation index, and obtains the evaluation result of the medical knowledge map.

The embodiment of the third aspect of the present application provides an electronic device, including a processor and a memory, where at least one instruction, at least one program, code set or instruction set is stored in the memory, and the instruction, the program, the The code set or the instruction set is loaded and executed by the processor to implement the steps of the medical knowledge map quality assessment system provided in the embodiment of the above-mentioned first aspect of the present application.

The embodiment of the fourth aspect of the present application provides a non-transitory computer-readable storage medium. When the instructions in the storage medium are executed by the processor of the mobile terminal, the mobile terminal can implement the above-mentioned first aspect of the present application. The steps of the medical knowledge map quality assessment system provided by the embodiment of one aspect.

The embodiment of the fifth aspect of the present application provides a computer program product. When the instructions in the computer program product are executed by the processor of the mobile terminal, the mobile terminal can execute the embodiment of the above-mentioned first aspect of the present application. Steps of the provided medical knowledge graph quality assessment system.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: the application discovers communities existing in different ontologies through pre-checking and the basic algorithm of community discovery in social networks, and utilizes the idea of community discovery in social networks to combine different ontologies Nodes with dense node relationships are condensed into communities through the network, and the quality of the graph network relationship between nodes is measured by the modified mutual information.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description serve to explain the principles of the invention.

Fig. 1 is a flow chart of a medical knowledge map quality assessment system according to an exemplary embodiment.

Fig. 2 is a block diagram of a medical knowledge map quality assessment device according to an exemplary embodiment.

Fig. 3 is an internal structural diagram of an electronic device according to an exemplary embodiment.

detailed description

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with aspects of the invention as recited in the appended claims.

Fig. 1 is a flow chart of a medical knowledge map quality assessment system shown according to an exemplary embodiment, as shown in Fig. 1, including the following steps:

In step S101, a medical knowledge map built based on the schema is acquired.

Specifically, the medical knowledge map is constructed based on the schema, and then the medical knowledge map is obtained, and the quality of the medical knowledge map is checked.

In step S102, a pre-check operation is performed on the medical knowledge graph, and the pre-check operation includes at least one of a defined field data type check, an outlier check, and a knowledge graph topology check.

Specifically, in order to make the subsequent social network detection more accurate, the medical knowledge map is checked by pre-checking the medical knowledge map by defining field data type checks, outlier checks, and knowledge map topology checks and fixes.

In some embodiments, the definition field data type check includes:

Based on the field type of the medical knowledge map, detect all fields of each entity;

Correct the abnormal result field, and record the proportion of all the abnormal result fields.

Specifically, according to the field types defined in the medical knowledge map, all fields of each entity are automatically detected, and the data structure of the fields detected as abnormal results is corrected. If it cannot be corrected directly, an abnormal prompt is performed, and manual correction is performed, and the proportion of all abnormal result fields is recorded as an indicator. In conjunction with specific embodiments, for measurable data types in data types, such as age, year, statistical data, ratio, etc., some of them need to define a value range when defining, such as age cannot be negative, maximum value setting, statistical data are all integers. When this situation is defined, the restrictions in the preset format will be written, and then a special program package will be written to analyze the conditions, so that after the path of the field is entered, the restrictions recognized by the update program can be obtained. When entering data through the update program, the update program will obtain the restrictions on the data types of these fields when reading the fields. If the data type or range is wrong, the updated data cannot be created normally, and the cause of the error and the correct type will be prompted example. For errors that can be automatically corrected, such as the data is a digital type of age, but the input is a string type, a forced type conversion will be performed. If it cannot be converted or the conversion still does not meet the restrictions, manual confirmation is required to modify it, such as age It is "28 years old" that cannot be forcibly converted into a numeric type. The unit of age has been set as "years old". You only need to write the specific number "28" of the age.

For non-quantifiable units, such as Boolean type, there are only two values of True and False. When manually editing this field, only these two options will be set, and there are no other options. Similarly, there are ethnicity, gender, etc., which can only Choose from preset options to avoid unexpected data type results.

For the above data type errors, the proportion of all abnormal result fields will be recorded as an evaluation indicator.

In some embodiments, the outlier check includes:

Outlier detection is performed on the point-in degree and point-out degree of each node in the medical knowledge graph, and the outlier points are merged into the graph structure of the medical knowledge graph.

Specifically, outliers are detected by using the in-degree and out-degree of each node in the medical knowledge graph, and the outliers are incorporated into the knowledge graph structure. In conjunction with specific embodiments, an outlier refers to an entity that is not associated with any other entity, and any relational triples in the map are missing. For data, the in-degree and out-degree of a node are Zero, the in-degree is the relationship that other entities point to this entity, that is, there is no such entity in the tail entity, and the out-degree is the relationship that this entity points to other entities, that is, the head entity does not have this entity. Through the traversal of all entities, entities that do not exist in the head and tail entities of the triples are detected, that is, outlier detection, and then all missing relationships need to be evaluated. For example, the treatment relationship of a disease needs to be supplemented. However, its complications do not necessarily need to be completed. Not all diseases have complications. After evaluating the relationship that needs to be completed, complete it through knowledge completion technology.

In some embodiments, merging the outlier into the medical knowledge graph structure includes:

The method of keyword matching and edit distance is used to recommend the parent node of the outlier point, and the outlier point is associated with the parent node of the outlier point.

Specifically, it is easy to know from the hierarchical structure of the graph that when including outliers in the graph structure, it is only necessary to associate them with their parent nodes. Therefore, using keyword matching and edit distance to recommend parent nodes makes outliers included into the knowledge graph structure. In combination with specific embodiments, since almost every entity in the medical graph has a relationship setting of a superordinate term during design, that is, a wider word is extended on the entity, so the superordinate term is usually completed first, This process is called outlier entry into the graph structure, so that other relationships can be inferred through this superordinate term. The superordinate term can be recommended by detecting the keywords in the entity name and combining its edit distance with other entity names, that is, the entity with a large similarity measure to the entity name that needs to be recommended is the recommended object. Record the proportion of outliers as a measure.

In some embodiments, the knowledge map topology inspection includes:

Directed acyclic graph structure detection using topological sorting;

Use DFS and graph connectivity detection to detect tree structures;

Automatic connection or disconnection processing for abnormal results.

Specifically, by using topological sorting to detect the directed acyclic graph structure of the medical knowledge graph, using DFS and graph connectivity to detect the tree structure of the medical knowledge graph, and automatically connecting or disconnecting the abnormal results.

In step S103, the medical knowledge map is checked based on social network community discovery and the amount of mutual information corrected as an evaluation index, and an evaluation result of the medical knowledge map is obtained.

In some embodiments, the amount of mutual information based on social network community discovery and correction is used as an evaluation index to check the medical knowledge map and obtain evaluation results, including:

Based on the basic algorithm in the social network, the communities existing in different ontologies are obtained, and the medical knowledge map is corrected by using the revised mutual information, so as to obtain the evaluation result of the medical knowledge map.

Specifically, after checking the outlier points and the topology of the knowledge map, the nodes in the two closely related ontologies are regarded as communities, and the community static discovery algorithm is used. After continuous network cohesion, indicators such as modularity are used. By comparing the connection density difference between the existing knowledge graph network and the benchmark network under the same community division, the quality of the graph network relationship between nodes is measured. Among them, Modularity is a commonly used standard to measure the quality of node grouping. The higher the modularity, the more the detected communities conform to the characteristics of "tight inside and loose outside", and the better the grouping quality.

The specific algorithm is as follows:

1. The greedy algorithm for maximizing modularity Fast NewMan (FN) calculates the static communities of A ontology and B ontology respectively, and keeps the number of communities k the same.

为A本体中第h和社区中节点数，设

为B本体中第h和社区中节点数。N表 示总节点数（A和B本体总和）。

为A本体中第h和社区中所有关系数，设

为B本体中第h和 社区中所有关系数。I表示总关系数（A和B本体总和）。 2. Set

is the hth node in the A ontology and the number of nodes in the community, set

It is the hth node in the B ontology and the number of nodes in the community. N represents the total number of nodes (sum of A and B ontologies).

is the hth in the ontology of A and the number of all relationships in the community, set

is the number of all relationships between the hth in the B ontology and the community. I represents the total relationship coefficient (sum of A and B ontology).

3. S is an association matrix, and the matrix indicates the number of external edges between the sub-communities in the two ontology community structures of A and B, as shown in Table 1.

Table 1

S11 … S1k … … … Sk1 … Skk

4. The amount of mutual information

，修正的MI值接近0 代表两个本体间关系独立，值越大代表两个本体间有越高的关联关系，对不同类别的事务 有越强的刻画能力。 5. Make corrections

, the modified MI value is close to 0, indicating that the relationship between the two ontologies is independent, and the larger the value, the higher the relationship between the two ontologies, and the stronger the ability to describe different types of transactions.

6. Select multiple ontologies and evaluate them sequentially. For the ontology with a lower corrected MI value, it is judged that its own objective correlation is poor, or it is caused by incomplete map construction, and then supplementary corrections are made.

There is a natural aggregation phenomenon in the relationship between entities in each ontology. For example, the relationship between upper respiratory tract infection diseases will be relatively close, such as viral influenza, pharyngitis, chronic pneumonia, and so on. And other diseases such as lower extremity osteoporosis and meniscus damage are relatively sparse. Using the community discovery algorithm of social network, diseases can be clustered spontaneously according to the degree of closeness. Other ontology will have the same situation. For example, cough, runny nose, sore throat, etc. will be closely related among entities of symptom ontology. Symptoms such as leg pain, knee soreness, lower limb weakness, etc. will be relatively close, and community Discover.

In terms of the relationship between symptoms and diseases, natural upper respiratory tract infection diseases are closely related to symptoms such as cough, runny nose, sore throat, etc., while diseases such as lower limb osteoporosis and meniscus damage are closely related to leg pain, knee soreness, lower limb pain, etc. There will be relatively more associations with symptoms such as weakness...

Mutual information can be used to measure the correlation of different communities between two types of ontologies.

和

的 权重修正。 At the same time, it is also considered that 1. When other conditions remain unchanged, the larger the community, the weaker the correlation. 2. When other conditions remain unchanged, the more relationships within the community, the weaker the correlation. Then proceed to the formula of mutual information

and

weight correction.

In summary, the medical knowledge map quality assessment system provided by this application, through the above steps, this application uses the basic algorithm of pre-checking and community discovery in social networks to discover communities that exist in different ontologies, and uses the idea of community discovery in social networks to Nodes with dense node relationships between different ontologies are condensed into communities through the network, and the quality of the graph network relationship between nodes is measured by the modified mutual information.

Fig. 2 is a block diagram of a medical knowledge map quality assessment device according to an exemplary embodiment. Referring to FIG. 2 , the device includes an acquisition module 201 , a pre-detection module 202 and a community network module 203 .

An acquisition module 201, configured to acquire a medical knowledge map built based on schema;

A pre-inspection module 202, configured to perform a pre-inspection operation on the medical knowledge graph, the pre-inspection operation at least including one of defined field data type inspection, outlier inspection, and knowledge graph topology inspection;

The community network module 203 checks the medical knowledge graph based on social network community discovery and corrected mutual information as an evaluation index, and obtains the evaluation result of the medical knowledge graph.

Regarding the apparatus in the foregoing embodiments, the specific manner in which each module executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

In one embodiment, an electronic device is provided. The electronic device may be a terminal, and its internal structure may be as shown in FIG. 3 . The electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Wherein, the processor of the electronic device is used to provide calculation and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, an operator network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a medical knowledge map quality evaluation system is realized. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball or a touch pad provided on the casing of the computer device , and can also be an external keyboard, touchpad, or mouse.

Those skilled in the art can understand that the structure shown in Figure 3 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation to the computer equipment on which the solution of the application is applied. The specific computer equipment can be More or fewer components than shown in the figures may be included, or some components may be combined, or have a different arrangement of components.

In one embodiment, the apparatus for assessing the quality of a medical knowledge graph provided in this application can be implemented in the form of a computer program, and the computer program can be run on an electronic device as shown in FIG. 3 . Each program module that constitutes the medical knowledge map quality assessment device can be stored in the memory of the electronic device.

At least one instruction, at least one section of program, code set or instruction set is stored in the memory in the electronic device, and the instruction, the program, the code set or the instruction set are loaded and executed by the processor to realize The medical knowledge map quality assessment system according to any one of the above embodiments. For example, implementing a medical knowledge map quality assessment system includes: obtaining a medical knowledge map built based on a schema; performing a pre-check operation on the medical knowledge map, and the pre-check operation includes at least defining field data type checks, outlier checks, knowledge One of graph topology checks; based on social network community discovery and corrected mutual information as an evaluation index to check the medical knowledge graph, and obtain the evaluation result of the medical knowledge graph.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: acquiring a medical knowledge graph built based on a schema; executing the medical knowledge graph on the Pre-check operation, the pre-check operation includes at least one of defined field data type check, outlier check, and knowledge map topology check; community discovery based on social network and the amount of mutual information corrected are used as evaluation indicators for all The medical knowledge map is checked, and the evaluation result of the medical knowledge map is obtained.

In one embodiment, a computer program product is provided. When the instructions in the computer program product are executed by the processor of the mobile terminal, the mobile terminal can perform the following steps: acquire a medical knowledge map built based on schema; The medical knowledge map performs a pre-check operation, and the pre-check operation includes at least one of defined field data type checks, outlier point checks, and knowledge map topology checks; the amount of mutual information based on social network community discovery and correction is used as The evaluation index checks the medical knowledge map to obtain the evaluation result of the medical knowledge map.

Those of ordinary skill in the art can understand that realizing all or part of the processes in the methods of the above embodiments can be completed by instructing related hardware through computer programs, and the computer programs can be stored in a non-volatile computer-readable storage medium , when the computer program is executed, it may include the procedures of the embodiments of the above-mentioned methods. Wherein, any reference to storage, database or other media used in the various embodiments provided in the present application may include at least one of non-volatile and volatile storage. Non-volatile memory may include read-only memory (Read-OnlyMemory, ROM), magnetic tape, floppy disk, flash memory or optical memory, and the like. The volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as Static Random Access Memory (SRAM) and Dynamic Random Access Memory (DRAM), among others.

The technical features of the above embodiments can be combined arbitrarily. To make the description concise, all possible combinations of the technical features of the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, all It should be regarded as the scope described in this specification.

The above examples only express several implementation modes of the present application, and the description thereof is relatively specific and detailed, but should not be construed as limiting the scope of the patent for the invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the scope of protection of the patent application should be based on the appended claims.

Claims (10)

1. A medical knowledge-graph quality assessment system, comprising:

acquiring a medical knowledge map constructed based on schema;

performing a pre-inspection operation on the medical knowledge-graph, wherein the pre-inspection operation at least comprises one of defining field data type inspection, outlier inspection and knowledge-graph topology inspection;

and checking the medical knowledge graph by taking the community discovered and corrected mutual information quantity of the social network as an evaluation index to obtain the evaluation result of the medical knowledge graph.

2. The medical knowledge-graph quality assessment system according to claim 1, wherein said defining field data type checks comprises:

detecting all fields of each entity based on the field types of the medical knowledge graph;

and correcting the abnormal result field, and recording the proportion of all the abnormal result fields.

3. The medical knowledge-map quality assessment system of claim 1, wherein said outlier inspection comprises:

and detecting outliers of the point-in degree and the point-out degree of each node of the medical knowledge graph, and combining the outliers into the spectrogram structure of the medical knowledge graph.

4. The medical knowledge-graph quality assessment system of claim 3, wherein incorporating the outliers into the medical knowledge-graph spectrogram structure comprises:

and performing parent node recommendation on the outliers in a key word matching and editing distance mode, and associating the outliers with the parent nodes of the outliers.

5. The medical knowledge-graph quality assessment system according to claim 1, wherein the knowledge-graph topology inspection comprises:

performing directed acyclic graph structure detection by using topological sorting;

detecting a tree structure by using DFS and graph connectivity detection;

and carrying out automatic connection or broken link processing on the abnormal result.

6. The medical knowledge-graph quality assessment system according to claim 1, wherein the medical knowledge-graph is checked to obtain an assessment result based on the community discovery of social networks and the corrected mutual information amount as an assessment index, and the method comprises:

and acquiring communities existing in different ontologies based on a basic algorithm in a social network, correcting the medical knowledge graph by using the corrected mutual information, and acquiring the evaluation result of the medical knowledge graph.

7. A medical knowledge-map quality assessment apparatus, comprising:

the acquisition module is used for acquiring a medical knowledge map built based on the schema;

the pre-inspection module is used for performing pre-inspection operation on the medical knowledge graph, and the pre-inspection operation at least comprises one of field data type inspection, outlier inspection and knowledge graph topological structure inspection;

and the community network module is used for checking the medical knowledge graph based on the community discovery and corrected mutual information quantity of the social network as an evaluation index to obtain the evaluation result of the medical knowledge graph.

8. An electronic device comprising a processor and a memory, the memory having stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, the instruction, the program, the set of codes, or the set of instructions being loaded and executed by the processor to implement the medical knowledge base quality assessment system according to any one of claims 1-6.

9. A non-transitory computer readable storage medium, wherein instructions in the storage medium, when executed by a processor of a mobile terminal, enable the mobile terminal to perform the medical knowledge graph quality assessment system according to any one of claims 1-6.

10. A computer program product, characterized in that instructions in the computer program product, when executed by a processor of a mobile terminal, enable the mobile terminal to execute the medical knowledge graph quality assessment system according to any of claims 1-6.

Images