Key points are not available for this paper at this time.
Abstract A distinction is made between bibliographical indexes for new and past literature based on the willingness of the user to trade perfection for currency. Indexes giving keywords in their context are proposed as suitable for disseminating new information. These can be entirely machine‐generated and hence kept up‐to‐date with the current literature. A compatible coding scheme to identify the indexed documents is also proposed. In it elements are automatically extracted from the usual identifiers of the document so that the coded identifier yields a maximum of information while remaining susceptible to normal methods of ordering.
H. P. Luhn (Sat,) studied this question.