The Information Content of Titles in Engineering Literature

Since many alerting and other information services rely very heavily on the use of titles to transfer information to the potential user, it is essential that he be aware of the proportion of the information contained in the complete document which will not be deducible from the title and which he will therefore miss. Methods will be discussed for analyzing the relative information content of the titles of engineering papers and results presented for the amount and type of information lost through scanning title listings only. Between one-third and one-half of indexable terms are not retrievable from article titles even if all possible synonyms and related terms are used. If all synonyms are used instead of one keyword, the amount of information retrieved is increased by about 70 percent. The problems of dealing with synonyms and with syntactical variants in searching titles indexes are discussed. The possibility of using keywords in journal titles as supplementary retrieval tags is suggested since they were deemed useful in nearly one-third of the sample of papers analyzed.