Searching
The Internet has provided an immense boost to the ability of computers to search databases for information on a myriad of topics. On-line searching enables the perusal of a variety of information sources. The explosive popularity of on-line searching is enhanced by the development of a number of search engines--programs that use a variety of mechanisms to search Internet sites for a requested topic, and then deliver the site to the user's computer.
Databases are often specific to one discipline, such as chemistry, physics, microbiology, or can be more general, including multiple disciplines to address themes such as energy or the environment. If a site has a number of databases and the appropriate tools, simultaneous searches across several databases is possible.
Bibliographic databases provide references to periodical articles and sometimes to other types of literature. Some of these databases provide abstracts, brief summaries, of the articles. Scientific and medical databases, such as MEDLINE, often carry article abstracts. Abstracts can be advantageous, allowing a rapid perusal of many articles, in order to focus a search to the truly germane articles. Full text databases provide the whole text of publications. A current trend is the user-pay domain site, where, for a membership or one-time fee, the user can download the text and graphics of an article. Databases that contain data are termed databanks. An example is the collection of Material Safety Data Sheets, a databank listing the properties of chemical compounds.
Aside from the search engines mentioned above, a number of search mechanisms exist. A popular search method is known as the keyword, or free-text search. This technique allows the search for the occurrence of words in any or some of the fields in a record. A popular key word search method is the Boolean search. A Boolean search operates based upon key words or phrases supplied by the user.
Variations on keyword searches aid the user in broadening or narrowing searches. Nesting is a variation where similar terms are placed in parentheses to specify the order in which the keywords occur. Another variation, called proximity, requires the keywords to appear within a certain number of words from each other. In effect, the words are tied together to direct the search. A related search is called adjacency. As its name implies, the keywords must appear adjacent to each other in the identified databases.
Searching can also be conducted using search trees. A search tree consists of so-called nodes. One is the root node, and the other nodes are partitioned into subtrees that are linked to the root node. If the search tree is either empty, or if each node has no more than two subtrees, than it is called a binary tree. A binary search algorithm is operative only if the data are sorted; the algorithm can then be useful to speed up a search. An instructive analogy is locating a name in a telephone book. Finding a name by sequentially searching from the beginning of the phone book is efficient only if the data entry is near the beginning. A more efficient strategy is to begin the search somewhere near where the user expects the entry to be. If the entry is not there, then the search will proceed in one direction or another. If the data is always split in half and the middle item checked, then the remaining items to be checked are halved every time. In contrast, a linear search can eliminate only one item at a time.
In a binary search, the above analogy is accomplished by the tree structure—a parent field, leftchild field and rightchild field. Fields at various positions allows a search of the database.
In contrast to a binary search, a sequential, or linear, search examines data items sequentially and one at a time until the target data item is identified. For example, a database could consist of a list of students and their corresponding grades. A search could be done using as the criterion the last name of a particular student. If the student exists, and their last name is in the list, the data point will be found and its corresponding grade will be retrieved. Searches can be limited to certain portions of a database, which is helpful when the databases are large.
Another way searches such as sequential searches can be made more efficient is by the use of a sentinel value. A sentinel value is a special value that a user assigns when the entry of a list of data is completed. Search algorithms incorporate the sentinel value--when the sentinel value is encountered the search is stopped. This allows a large database to be divided into smaller, more easily searched sections.
Hashing, another search technique, produces hash values. A hash value, or hash, is a number generated from a string of text by a formula. The number is unique; it is very unlikely that another text would produce the same hash value. Hashing enables a quick location of a data record after all the records have been treated with the formula to generate hash values for each data record. Hashes also play a role in security systems, ensuring that transmitted messages have not been tampered with. Comparison of hash values in the original message and the received message is accomplished. If both sets of hash values are the same, there is a high probability that the message was not altered during transmission.
This is the complete article, containing 893 words
(approx. 3 pages at 300 words per page).
