Text databases search techniques

In the simplest sense, searching chemical information consists of (IXinding structures or reactions that meet the chemist s search criteria and/or (2) finding data that meets the search criteria. Data searching (numbers and text) is a well-established informatics activity, supported by spreadsheets, word processors, and relational database systems. Chemical structures and reactions are a unique form of data. Searching for full or partial matches to structures, models, and reactions requires highly specialized databases and search techniques. 

Searching of one or more on-line databases is a technique increasingly used ia novelty studies. The use of such databases enables the searcher to combine indexing parameters, including national and international classifications natural language words ia the full text of patents, ia their claims, or ia abstracts suppHed by iaventor and by professional documentation services and indexing systems of various sorts. Because the various patent databases have strengths and weaknesses that complement each other, the use of multiple databases is thus pmdent, and is faciUtated by multifile and cross-file techniques provided by the various on-line hosts. 

A number of methods have been developed to introduce context to on-line databases, enabling searches to be refined to minimized false retrieval. One of the earliest techniques is proximity searching, in which two words are required to be adjacent, or within a limited distance from each other in text. The assignment of roles to chemical substances is a method of precoordinating concepts. A substance can be identified as a reactant, as a product, and in some systems in a number of additional roles. For example, by searching for documents in which formaldehyde is a product, documents in which it is a reactant, or in which it undergoes no reaction, are thus eliminated. 

Full-Text Patent Databases. The LEXPAT database on the LEXIS— NEXIS system, the first commercially available full-text patent file, receives its greatest use from patent attorneys and has been relatively unused by other patent information speciaUsts. This may be attributed to search software that is quite different from the type familiar to information speciaUsts, no matter what their preferred host system. This situation has changed with dialog s release of the PATEULL files followed by STN s USPATEULL, both searchable by familiar Boolean techniques and featuring greater... 

Our approach to this chapter is best described as experimental rather than mathematical. There are many excellent formal texts, some of which will be cited. Here, we wish to summarize the fundamental purpose of each technique, to explore its possible applications and limitations, and to illustrate (or provide references to) its use in systematic studies of molecular structure. Most of the examples have been drawn from the Cambridge Structural Database [2], using the methods of search and retrieval detailed in the preceding chapter. We preface the statistical content with two more general chemical sections in the first we discuss the selection of geometric parameters that are most appropriate for certain types of analysis, whilst in the second we discuss the possible sources of variation in crystallographic structural data. 

---