Loading...
Please wait, while we are loading the content...
Similar Documents
The Next Twenty Years in Information Retrieval : Some Goals and Predictions
| Content Provider | Semantic Scholar |
|---|---|
| Copyright Year | 2010 |
| Abstract | THIS discussion concerns a mechanized information retrieval system for the technical library of General Electric's Aircraft Gas Turbine Division. The system is confined to the technical reports and papers available in the Division's Library. Textbooks have not been included as of yet since they are carried on a Library of Congress system and did not readily fit into the manual scheme which was developed. This discussion will be in two parts. The first part will cover the information retrieval system prior to mechanization, and the second part will cover the mechanization of this system. The technical library was established in 1953 using a uniterm or key word coordinate indexing system. To understand this system, we will follow the progress of a publication through the various steps of the system. First, as a publication is entered into the library, it is assigned a six-digit access number (see Fig. 1). This report is typical of the technical reports generated within the Division. Another example of reports in the library would be NACA technical reports. The next step in this system is the abstracting of the document. The abstracting is done by professional librarians and then posted to a card file as shown in Fig. 2. This card is controlled according to the previously assigned access number. The next step in the system consists of reviewing the title, abstract, and document to select the most descriptive words which will identify the document. These words are primarily nouns. They become the uniterms. In our hypothetical case, these words are shown on the right in Fig. 2, just as they appear in the system. These uniterms, along with the appropriate access number, are then posted to the uniterm file (see Fig. 3). There are 100 numbers per side of card when full. Both sides are used, and in the case of general terms such as these, they are heavily posted so that several cards are required. Certainly this system appears cumbersome at this point, but it has the advantage that, in any given technical area, the number of uniterms tends to level off at a specific number after the system is developed. In our case, this number is something under 9000. The combined system is shown schematically in Fig. 4. I think now you can see how information is recalled from this system. The requestor discusses his problem with the librarian. They decide on the uniterms to search. The librarian then furnishes the appropriate uniterm cards to the requestor. Once again referring to Fig. 3, the problem facing him can be seen. He must cross coordinate the cards to find numbers which apply to all uniterms. In our case, we have used three uniterms. A |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://csdl.computer.org/csdl/proceedings/afips/1959/5054/00/50540077.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
