Showing papers on "Ranking (information retrieval) published in 1973"

Recent Studies in Automatic Text Analysis and Document Retrieval

Abstract: Many experts in mechanized text processing now agree that useful automatic language analysis procedures are largely unavailable and that the existing linguistic methodologies generally produce disappointing results. An attempt is made in the present study to identify those automatic procedures which appear most effective as a replacement for the missing language analysis.A series of computer experiments is described, designed to simulate a conventional document retrieval environment. It is found that a simple duplication, by automatic means, of the standard, manual document indexing and retrieval operations will not produce acceptable output results. New mechanized approaches to document handling are proposed, including document ranking methods, automatic dictionary and word list generation, and user feedback searches. It is shown that the fully automatic methodology is superior in effectiveness to the conventional procedures in normal use.

Ranking Characters by a Dispersion Criterion

Abstract: IT is often proposed by taxonomists and ecologists that classifications and other related analyses should be based on the possible largest number of characters. They believe that when large character sets are used the analysis will gain in accuracy, and the results will be more broadly applicable. Yet sampling, computational and other logistic problems are often associated with the handling of large character sets. Some characters may have to be deleted in the interest of more economical sampling and computations.

Ranking in incomplete blocks. The use of the technique and the interpretation of the results

Abstract: It is often more convenient or expedient to obtain data in the form of rankings than in the form of scores or measurements. Analysis of such data has, however, been difficult to perform and to interpret except for the most simple experiments. The model described by Bradley and Terry1 provides a very useful analysis in the case of ranking by paired comparisons. A method of analysis similar to the Bradley-Terry model, but dealing with ranking in incomplete blocks or complete blocks of any size, has been developed. The results of this analysis are readily interpretable and should enable more use to be made of ranking procedures.

Measuring English Teachers' Changes in Values Toward "Growth Through English" Variables.

Abstract: "Growth Through English" Variables Teacher trainers have long wanted to know how effective they were in modifying the attitudes of their students. In this study an attitude measure dealing with the "Growth Through English" philosophy, as described in the reports of the Dartmouth Seminar, is developed and tried out. The results indicate that a simple ranking of values is almost as effective as an elaborate forced-choice measure, and that a workshop can modify the expressed attitudes of teacher-participants.Reviewed by T. D.

A Note on Project Selection in Research and Development — Post-Rothschild

Abstract: THE PROJECT selection problem has been tackled by numerous people, including many from operational research backgrounds. The methods suggested and adopted include both qualitative and quantitative, the latter ranging from simple ranking procedures for individual projects, occasionally making use of decision-tree type structures, to those that consider a portfolio of projects and which often take the form of mathematical programming models. Information inputs are generally in the form of estimates of resource requirements and cash flows over time. Ranking is usually done in order of net present value, rates of return or benefit to cost ratios. In the portfolio type of models it is possible to choose a set of projects which maximize a given objective function subject to meeting a specified set of constraints or alternatively it is possible to use the model to identify a range of portfolios with particular characteristics from which an appropriate choice can be made. But what of recent developments in thinking about R & D management? In the applied research area, the recent Rothschild report' emphasizes the role of the "customer" in the project selection process and states that "the person who has to pay for the end-product of applied R & D is, for obvious reasons, in the best position to decide how much can be afforded". It goes on to say that "depending on his qualifications, he may need to take the advice of scientists, mathematicians or engineers skilled in the subject under consideration" but the inference is that the decision to go ahead or not with a particular piece of applied research ultimately lies with the "customer". If this is the situation we face we must ask on what basis will he make this choice. Presumably the answer is on two major points firstly the price he must pay and secondly the time-scale within which he can expect results in effect the "quotation" he receives from the R & D department or establishment. It would, of course, be naive to assume that the "customer-contractor" relationship is a simple one in which quotations can be prepared and agreed with the minimum of consultation. In areas of less complexity than R & D it is accepted that interchange takes place during the course of tendering but in many of these cases the customer knows as much as the contractor about the problems and difficulties which are likely to be encountered. This is not always true of R & D, which tends to offer a specialist service and where the ground which is common to both parties may be very limited. Indeed one of the

Paired comparison ranking by linear programming

Abstract: The problem of estimating partial and complete paired comparison rankings is formulated as a linear programming problem. The solution of the problem immediately yields a ranking. A treatment of ties is considered as a result of which the solution of the program may explicitly have two objects tied in the ranking. Such a ranking is called partial. A method for resolving ties and thus obtaining a complete ranking is given based upon the solution of the linear program. A numerical example is given to illustrate the subject matter.