UMLS语义网络详解

来源：
https://www.ncbi.nlm.nih.gov/books/NBK9679/

Semantic Network

The Semantic Network consists of (1) a set of broad subject categories, or Semantic Types, that provide a consistent categorization of all concepts represented in the UMLS Metathesaurus, and (2) a set of useful and important relationships, or Semantic Relations, that exist between Semantic Types. This section of the documentation provides an overview of the Semantic Network, and describes the files of the Semantic Network. Sample records illustrate structure and content of these files.

The Semantic Network is distributed as one of the UMLS Knowledge Sources and as an open source resource available on the Semantic Network Web site.

1.overview

The purpose of the Semantic Network is to provide a consistent categorization of all concepts represented in the UMLS Metathesaurus and to provide a set of useful relationships between these concepts. All information about specific concepts is found in the Metathesaurus. The Network provides information about the set of basic semantic types, or categories, which may be assigned to these concepts, and it defines the set of relationships that may hold between the semantic types. The Semantic Network contains 127 semantic types and 54 relationships. The Semantic Network serves as an authority for the semantic types that are assigned to concepts in the Metathesaurus. The Network defines these types, both with textual descriptions and by means of the information inherent in its hierarchies.

The semantic types are the nodes in the Network, and the relationships between them are the links. There are major groupings of semantic types for organisms, anatomical structures, biologic function, chemicals, events, physical objects, and concepts or ideas. The current scope of the UMLS semantic types is quite broad, allowing for the semantic categorization of a wide range of terminology in multiple domains.

The Metathesaurus consists of terms from its source vocabularies. The meaning of each term is defined by its source, explicitly by definition or annotation; by context (its place in a hierarchy); by synonyms and other stated relationships between terms; and by its usage in description, classification, or indexing. Each Metathesaurus concept is assigned at least one semantic type. In all cases, the most specific semantic type available in the hierarchy is assigned to the concept. For example, the concept "Macaca" receives the semantic type "Mammal" because there is not a more specific type "Primate" available in the Network. The level of granularity varies across the Network. This has important implications for interpreting the meaning (i.e., semantic type) that has been assigned to a Metathesaurus concept. For example, a sub-tree under the node "Physical Object" is "Manufactured Object". It has only two child nodes, "Medical Device" and "Research Device". It is clear that there are manufactured objects other than medical devices and research devices. Rather than proliferate the number of semantic types to encompass multiple additional subcategories for these objects, concepts that are neither medical devices nor research devices are simply assigned the more general semantic type "Manufactured Object".

Figure 1 illustrates a portion of the Network. The semantic type "Biologic Function" has two children, "Physiologic Function" and "Pathologic Function", and each of these in turn has several children and grandchildren. Each child in the hierarchy is linked to its parent by the "isa" link.

[图片上传失败...(image-384500-1572765886065)]

Figure 1.

A Portion of the UMLS Semantic Network: “Biologic Function” Hierarchy

The primary link in the Network is the "isa" link. This establishes the hierarchy of types within the Network and is used for deciding on the most specific semantic type available for assignment to a Metathesaurus concept. In addition, a set of non-hierarchical relations between the types has been identified. These are grouped into five major categories, which are themselves relations: "physically related to", "spatially related to", "temporally related to", "functionally related to", and "conceptually related to".

Figure 2 illustrates a portion of the hierarchy for Network relationships. The "affects" relationship, one of several functional relationships, has six children, including "manages", "treats", and "prevents".

[图片上传失败...(image-5c42ed-1572765886065)]

Figure 2.

A Portion of the UMLS Semantic Network: “affects” Hierarchy

The relations are stated between high level semantic types in the Network whenever possible and are generally inherited via the "isa" link by all the children of those types. Thus, for example, the relation "process of" is stated to hold between the semantic types "Biologic Function" and "Organism". Therefore, it also holds between "Organ or Tissue Function" (which is a "Physiologic Function", which is, in turn, a "Biologic Function") and "Animal" (which is an "Organism"). The relations are stated between semantic types and do not necessarily apply to all instances of concepts that have been assigned to those semantic types. That is, the relation may or may not hold between any particular pair of concepts. So, though the relation "evaluation of" holds between the semantic types "Sign" and "Organism Attribute", a particular sign or a particular attribute may not be linked by this relation. Thus, signs such as "overweight" and "fever" are evaluations of the organism attributes "body weight" and "body temperature", respectively. However, "overweight" is not an evaluation of "body temperature", and "fever" is not an evaluation of "body weight".

In some cases there will be a conflict between the placement of types in the Network and the link to be inherited. If so, the inheritance of the link is said to be blocked. For example, by inheritance, the type "Mental Process" would be "process of" "Plant". Since plants are not sentient beings, this link is explicitly blocked. In other cases the nature of the relation is such that it should not be inherited by the children of the types that it links. In that case, the relation is defined for the two semantic types it explicitly links, but blocked for all the children of those types. For example, "conceptual part of" links "Body System" and "Fully Formed Anatomical Structure", but it should not link "Body System" to all the children of "Fully Formed Anatomical Structure", such as "Cell" or "Tissue".

Several portions of the MeSH hierarchy have been labeled with child to parent semantic relationships. All of the anatomy, diseases, and psychiatry and psychology sections have been labeled, as well as a portion of the biological sciences section. The links that are expressed between MeSH terms are, with a few exceptions, reflected in the Semantic Network. That is, if two MeSH terms are linked by a certain relation, then that link is expressed in the Network as a link between the semantic types that have been assigned to those MeSH terms. For example, "Amniotic Fluid", which is a "Body Substance", is a child of "Embryo", which is an "Embryonic Structure". The labeled relationship between "Amniotic Fluid" and its parent "Embryo" is "surrounds". This is allowable, since the relation "Body Substance surrounds Embryonic Structure" is represented in the Network.

Figure 3 shows a portion of the Semantic Network, illustrating the relations, either hierarchical or associative, that exist between semantic types.

[图片上传失败...(image-444c29-1572765886065)]

Figure 3.

A Portion of the UMLS Semantic Network: Relations

The UMLS Semantic Network is provided in two formats: a relational table format and a unit record format.

2.Semantic Network ASCII Relational Format

There are two basic tables, two ancillary tables, and two bookkeeping tables included in this format. The two basic tables contain exactly the same information as the unit record file, but the information is presented differently. One table contains definitional information about the semantic types and relations; the other contains information about the structure of the Network. Each semantic type and each relation has been assigned a four character unique identifier (UI). These are of the form "T001", "T002", etc. The ancillary tables are expansions of the table that contains the Network structure. They give the fully inherited set of links represented in the Network. The first table is expressed as triples of UI's. The second is expressed as triples of names. The two bookkeeping tables describe the relational files and their fields. Fields in all tables are separated by a "|". All tables are listed and described below:

Table	Description
SRDEF	Basic information about the Semantic Types and Relations.
SRSTR	Structure of the Network.
SRSTRE1	Fully inherited set of Relations (UI's).
SRSTRE2	Fully inherited set of Relations (names).
SRFIL	Description of each table.
SRFLD	Description of each field and the table(s) in which it is found.