Proceedings of the International Conference on Scientific Information Two Volumes (1959) / Chapter Skim
Currently Skimming:
The Basic Types of Information Tasks and Some Methods of Their Solution
The Basic Types of Information Tasks and Some Methods of Their Solution
Pages 823-854
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 823... ...
On the basis of considerations which will be set forth below, the universe E is, as a rule, made up of homogeneous ek. Proceeding from the defined set of information, the scope of the questions asked is also determined more or less accurately.
|
From page 824... ...
, it is obviously desirable either to record, by some method, the results of searches previously made, or to create a special information language and to use it to record the content of the information in all the en' as well as the content of the questions asked, in such a form that will make it possible, to a greater or lesser extent, to automatize the process of solution, without at the same time leading to an inadmissibly large increase in the volume of initial data or time expended for the solutions. The first method of solution cannot be applied to all possible questions (of the clefined scope)
|
From page 825... ...
It is on the basis of these degrees of approximation that the basic types of information tasks will be introduced.
|
From page 826... ...
Since it is assumed that ek is taken as the sets of information which are not able to be subdivided into completely unrelated parts, the parts within the formula are either related by simple conjunctive relationships (type III) or by synthetic relationships (type IV)
|
From page 827... ...
Below we shaD consider information language and types of information tasks from a more general point or view. CONDITIONS FOR SELECTION The conditions for selection evolve from the ordinary posing of the task of searching for information, when it is necessary to find to collect information devoted to each of the "subjects" making up the volume of the concept characterizing the question asked.
|
From page 828... ...
INFORMATION LANGUAGE The characteristics An' which form the universe P and synthetic relations are analogous to the words and grammatical relations of ordinary language.
|
From page 829... ...
Then choice is made of the most appropriate level of concreteness of the independent characteristics forming the universe P With the aid of the generally used method of approximation, the characteristics from P must describe the content of ek and the questions with an accuracy to any permissible degree of concreteness.
|
From page 830... ...
separately from the basic task when approximating the question asked, if only this cloes not result in a too complicated representation of the question, for example, in the form of listing a large number of characteristics revealing the volume of less concrete characteristics; (b) in the process of solution, when, for the content of the ek being examiners more general representations of it than that which is recorded are found very quickly in a special memory device and all the representations are compared with the question, which is approximated, like the content of en in instance (a)
|
From page 831... ...
Let us note that the first two methods of solving the auxiliary information task presuppose that the transition to approximation of the content of ek at a lower level of concreteness of characteristics can be made by means of representation of each individual characteristic through less concrete characteristics. In the third method of solution this cannot be presupposed; however, then this approximation will be made too labor-consuming and nonautomatic and will require a still greater increase in the number of entries.
|
From page 832... ...
An analogous assumption can also be made with respect to synthetic relations. The types of information tasks revealed below by this assumption will obviously also occur for auxiliary information tasks, which, if they are viewed separately, do not in any way slider in principle from the basic tasks (some of the characteristics of the basic task begin to play the role of elements of information in the auxiliary task)
|
From page 833... ...
TYPE II. In the second type of information tasks each ek is characterized not by one standard name, which cannot be the case, but more thoroughly by several sections—by one or several standard characteristics in each section, with several characteristics in one section being related purely accidentally, as formerly.
|
From page 834... ...
If the latter occurs for all sections, then obviously, auxiliary searches to determine analytically related characteristics will, as before, not be necessary. Information tasks of the second type are a general instance with respect to tasks of the first type and are encounterer!
|
From page 835... ...
_ ~ .. ~ ~ The fourth type of tasks diners trom the second and third type by the fact that for elements of information and the questions asked, not only the characteristics are indicated, but also the synthetic relations between them within these elements and the questions.
|
From page 836... ...
Let us note that when searching for structures the distinction between characteristics and relations is beginning to become lost, and sometimes it proves more profitable to change over to a dual representation of mutually related characteristics, assuming the relations to be characteristics, and vice versa. The questions in the fourth type of tasks consist, as a rule, of mutually related characteristics.
|
From page 837... ...
The latter task is encountered especially frequently when making a combined solution, by the third method, of a basic and auxiliary information task, when the dependent, more concrete characteristics are represented by less concrete characteristics. Coding CODE DESIGNATIONS AND SYMBOLS Before describing some methods of solving information tasks of the types introduced above, we should like to discuss a few general considerations regarding the coding of standard characteristics and synthetic relations.
|
From page 838... ...
Whereas during nonautomaticady controlled searches the characteristics and synthetic relations are recorded in both approximations in the usual standard form—in the form of words, numbers' or other symbols during automatic searches use is made not of this form, but of code designations consisting of groups of code symbols situated one way or another on the medium and in the memory. Code symbols can be, for example, notches or perforations on punches!
|
From page 839... ...
3. Nonlocal code, when one does not know beforehand the location of the fields set aside for recording the individual characteristics, as is typical for type III and type IV tasks.
|
From page 840... ...
The third group of demands made upon coding evolves from the demand of simplicity and automatic control in finding the code values themselves. In general the finding of the code designations of characteristics according to their standard characteristics represents a type ~ information task and can be arhi.
|
From page 841... ...
With the simultaneous solution of the basic and auxiliary information tasks by the third method, analogous demands also occur with respect to the coding of the dependent characteristics, since the less concrete characteristics approximate such more concrete characteristics for one of the four types considered above. With the aid of fortuitous coding it is sometimes possible to avoid a considerable increase in the volume of recording each dependent characteristic, if one reduces its former code designation to a combination of code designations approximating its less concrete characteristics.
|
From page 842... ...
Some methods of solving information tasks TYPE I TASKS To solve such tasks, most often one uses ordinary dictionaries, reference books, encyclopedias, printed subject indexes (with standard subject headings) , tables, card files, and the like.
|
From page 843... ...
. Consecutive-action decoders, which use punched cards, punched tape, magnetic tape, photographic film, etc., are only partially free of the shortcomings mentioned and, although they do possess the proper automation of searches, in the case of medium numbers of ek, despite .
|
From page 844... ...
, it is not possible here to limit oneself only to the consecutive isolation of subsets inserted in one another, but it is necessary also to search for the intersection of rather large subsets from E which correspond to the individual characteristics of the question. And this does not make it possible to use, in the case of type II tasks, certain of the simpler parailelaction decoders, as, for example, the Amfis system.
|
From page 845... ...
where Pk C P is a set of characteristics approximating the corresponding ek, and In ~ P* , and let us assume that the code designations having the identical number of bits, the designations of In on the medium and of Pn in the memory, contain an identical number of code symbols (represented in the form 1 on the medium and in the form 1 in the memory)
|
From page 846... ...
. In the case of superimposed punched cards, ~ is taken as the punched-out cell, is the superimposed card, and ~ is the passage of light rh~c~u~h the re11 writh .1 1 r 1 ~ ~ I___ _,,, ~ ~^ Ace_ ~~~ VV 1~1 tne absence ot an obstacle in that cell on the superimposed punched cards (M1 n Mo = 0)
|
From page 847... ...
. For medium numbers of ek, the described decoders produced have as yet, unfortunately, been too complicated, and this makes it possible in this instance to attempt to improve immediately the system of superimposed punched cards, where the finding ot the Intersection or seas mn'' mn2' - ~ is R11Ccessfu~v made automatic, requiring, however, the recording of each set according to a direct code, that is, on each card (or set of cards)
|
From page 848... ...
The way out of this position must apparently be sought (if one does not return again to complex nonmechanical decoders) in rein the Air ~~^ ~~- I_; ~ r~ ~ ~ ~ J ~ __O -A ~^,-~ aver Al 1~l~llls 5~5 pal' Any' ~ anal and in recording the answer En for each characteristic an according to a nonlocal code on a group of punched cards or on a piece of punched tape, photographic film, or the like, in the form of a sequence of code designations of numbers of ad ek ~ En.
|
From page 849... ...
TYPE III TASKS ~ ~ . In order to solve these tasks one may use the parallel-action decoders described in the preceding section, only in the instance that the characteristics of Pn from the set Pk which describes the corresponding element of information 1 ~7 ek' are recorded by direct code, that Is, each In ~ P is given its own place in the simplest "devices," its separate card in a system of superimposed punched cards, its cell on punched cards with notches on the edges, and its input line in electrical decoders.
|
From page 850... ...
Sectors of the medium which contain the code designations of the characteristics from Pk which are recorded by a nonlocal code are consecutively passed or switched up to the scanning device of the decoder. Then each In from the Pk is compared with all the In from the set of question characteristics P*
|
From page 851... ...
Thus, with medium numbers of ek the devices that can be considered the best adapted for the automatic solution of type III tasks are consecutive-action decoders or an improved decoder with superimposed punched cards and with the use of a superpositional code, but with large numbers of ek the best device is the improved parallel-consecutive-action device which was described at the end of the preceding section. TYPE IV TASKS Generally speaking, only consecutive-action decoders can be employed for the solution of these tasks, since the synthetic relations do not characterize directly the ek themselves, but are built up on characteristics from ek.
|
From page 852... ...
For medium numbers of ek the automatic solution of a type III task (with the aid of a parallel-action crevice) produces, as we have seen, only the numbers of these ek, and therefore the locating by numerical designations of a rather large number of complete descriptions ofthese ek, as in type I tasks, can occupy too much time for subsequent searches taking the relations into consideration.
|
From page 853... ...
W PERRY, Punched Cards Their Applications to Science and!
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.