A database is a collection of interrelated data reasonably organized in a certain way and stored on computer storage devices. It is the product of the combination of computer technology and information retrieval technology, the main body of electronic information resources and one of the core parts of information retrieval system. According to the information provided, databases can be mainly divided into reference databases and source databases. 1. Reference database is a kind of database which mainly stores a series of descriptive information content and guides users to another information source to obtain complete original information, mainly including bibliography database and guide database. (1) Bibliographic database is a database for storing bibliographic clues such as contents, titles and abstracts. Also known as the secondary literature information database. For example, various library catalog databases, bibliographic databases, abstract databases, etc. all belong to this category, and their function is to point out the clues for users to obtain original information. Library catalog database, also known as machine-readable catalog, has detailed data content, which not only describes bibliographic information such as title, author, publishing project, but also provides collection information for users to request original information. The data content described by the title and abstract database is similar to that of printed titles and abstracts, which provides certain information sources such as paper information or patent information for users to retrieve. (2) Guide database stores a database describing objects about institutions, people, products, activities, etc. Compared with other databases, the guide database not only provides users with relevant information, but also includes various types of entities, and most of them use names for retrieval. For example, the directory database of institutions that store information about production and business activities, biographical database that stores information about personnel, product guide database that stores information about products or commodities, and fund database that stores information about funds all belong to this category, and its function is to guide users to obtain more detailed information from other relevant information sources. 2. The source database mainly stores full text, numerical values, structural information, etc., and can directly provide original information or specific data, and users do not need to refer to the databases of other information sources. It mainly includes full-text database and numerical database. (1) The full-text database is the source database that stores the full text or main part of the original information. Such as periodical full-text database, patent full-text database, encyclopedia full-text database, etc., users can directly retrieve the full text containing the original information of a word or phrase. (2) The numerical database stores a source database that mainly represents numerical information, and there is a similar text-numerical database. Compared with bibliographic database, digital database is the product of deep processing of information, which can directly provide needed data information. Such as various statistical databases, scientific and technological databases, etc. In addition to the general retrieval function, the numerical database also has the functions of accurate data operation, data analysis, graphic processing and sorting and reorganizing the retrieved data. 4.2.2 Database structure 1. Bibliographic database structure Bibliographic database organizes a series of data in the form of documents, which are called records, and a record contains several fields. (1) records and field records are a collection of related data processed as a unit, and they are the basic data units that make up a document. Several fields contained in the record are the basic data units that make up the record. In the bibliographic database, a record is equivalent to a title or abstract, so a record usually consists of title field, author field, source field, abstract field, subject field, classification number field, language field and so on. In some fields, there are several subfields, and the subfield is the lower data unit of the field. For example, the Subject field contains multiple subject words. According to the nature of the records represented by fields, fields are usually divided into two categories: basic fields and auxiliary fields. See Table 4- 1 for common field names and codes. Table 4- 1 field name and code basic field auxiliary field field name field name field name field code title Ti record number DN abstract AB author Au narration de author unit CS identification word ID periodical name JN publication year PY publication country CO language LA(2) A collection of related records organized according to a certain structure. Literature is the basic form of data organization in bibliographic database, and the way of literature organization is closely related to the software and hardware functions of retrieval system. In the bibliographic database, the literature structure is mainly divided into in-line literature and reverse literature. 1) Inline document records are stored in order, and the logical order between records is consistent with the physical order, which is equivalent to the arrangement order of abstracts in printing tools, and is a linear document. Inline document is the main part of the database, but the identification of its subject words and other features is out of order. When searching directly, you need to search the complete records from beginning to end, which takes a long time and is not practical. 2) The reverse document extracts fields with theme properties (such as theme field, title field, description field, etc.). ) and fields without discipline nature (such as author field, institution field, source field, etc.). ) and reorganize them in a certain order to get a document. The reverse gear with theme nature is called the basic index file, and the reverse gear without theme nature is called the auxiliary index file. To sum up, the main difference between forward shift and backward shift is that forward shift is the main document with the complete record as the processing and retrieval unit, while backward shift is the index document with the field in the record as the processing and retrieval unit. When searching by computer, first enter reverse gear to find the reference number of relevant information, and then enter forward gear to find records according to the reference number. 2. The structure of full-text database Generally, the structure of full-text database is similar to that of bibliographic database. The record in the full-text database is full-text text, which is divided into several fields. Its main document is a text document organized in the form of forward lines, and the reverse document is an index document corresponding to the detectable fields of records. 3. The structure of the numerical database The structure of the numerical database should comprehensively consider the content and retrieval purpose of the database, that is, in terms of content, the main content of the numerical database is numerical information, but the necessary descriptive text information is not excluded. In retrieval, it is not only convenient for single retrieval, but also convenient for comprehensive retrieval, and it can also perform accurate data operation, data analysis, graphic processing and sorting and reorganization on the retrieved data. The data structure of a digital database can be in the form of cells or lists. The former is the simulation of original data, and the latter is the machine-readable simulation of statistical tables. A digital database usually has various documents, such as embedded files, inline files, index files and so on. Forward movement consists of numerical data, which is the main document and the corresponding index document. In order to facilitate access, index files are organized in the form of basic direct access structure. Reverse gears also have corresponding index documents, which are organized hierarchically. The document structure of numerical database makes all documents available for retrieval and all data available for calculation, which constitutes the characteristics of numerical database. 4. Structure of the guide database The structure of the guide database has the characteristics of bibliographic database, full-text database and digital database, including forward file, reverse file, index file and data dictionary. Generally speaking, for a large-scale guide database with many disciplines and comprehensive contents, the main document can adopt the structure of multiple subdocuments, while for a single discipline and specialized content, it is appropriate to adopt a single main document and a variable-length multi-field recording format.