JPS6191723A - Sequence monitor and classification processing system - Google Patents

Abstract

PURPOSE:To shorten a processing time without reference to the arrangement of input data by changing the sequence of string generation or merging plural strings when the length of a string is longer than a specific length. CONSTITUTION:A string generating means 1 consists of a substitution and selection part 1a and a comparison part 1b; the substitution and selection part 1a constitutes a tournament tree on its internal storage element and determines winners and losers at respective nodes of the tree, and the comparison part 1b makes a record comparison on the basis of a classification key at each node. A string length monitoring means 2 detects the length of a string outputted to a string storage means 3 and compares it with a reference value; when it is smaller than the reference value, the comparison part 1b inverts the comparison of the next string of the tournament to generate a string in the reverse sequence. A string merging means 4 when all data are inputted reads strings out of the string storage means 3 and beings to merge them, the merged string is rewritten in the string storage means 3 newly, and the merging operation is repeated.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a classification processing method for office data based on a city, a sub-institutional district, etc. 0 is a sequence of data arranged to some extent from input data (hereinafter referred to as This invention relates to an order-monitoring classification processing method that generates and merges strings (referred to as strings).

[Conventional technology]

Data classification processing is frequently performed in the office processing field.
[Two high and far processing is always required. In the conventional classification processing method, as shown in FIG. 4C-2, it is generally composed of six 7-A's: incoming cuff needs 11, merging phase 12, and out cuffing needs 13. In the input phase 11, the input data is read, the internal storage creates a string in the working area above, and this string is outputted to the external storage, and after inputting all the data, the merge phase 12 stores these. The merging of the strings is repeated, and after the final merging, the data is output in an output phase 16.

[Problems that the invention attempts to resolve]

For high-speed classification processing, it is essential to shorten the time required for merging 7Azes, but the time required for merging processing (the number of strings created by Ma input Cuff Needs 11) In the conventional method described above, the number of strings created in the manual phase 11 is usually determined by the sequence of input data (binary 1f, In particular, there is a drawback that many strings are generated for input data sequences that are in the opposite order to the order in which they should be classified, and that it takes a long time to complete the process. [Means for] The present invention (the order monitoring classification processing method according to 2) includes a means for generating a string, a means for monitoring the length of the generated string, and a means for monitoring the length of the string from ≦2. If the accuracy is lower than a predetermined value, a means for changing the generation order of the next string, a means for merging the generated strings, and a classification process are performed in order to shorten the processing time without depending on the input data arrangement. It was designed to make the

[Good example]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows one example of the order monitoring classification processing method according to the present invention C.
This is a block color indicating G 9i. An example of this is string generation means 1, string length monitoring means 2, string storage! +I:? 3, and string merging means 4.

The string generation means 1 compares the string change selection section 1a with the bitl
It consists of b. In the selection section 1a, a tournament tree as shown in section 2 of Figure 2 is created on the internal storage element, and winners and losers at each node C'' of the tree are determined.In the comparison section 1b, , records are compared using the molecule J+key at each node.

The specific flow is shown in the 21st (2nd section).The winner at the top node of the tournament tree
U''') are output and form a string.

Then, the tlDt secondary input record (1T” in Uη(2)) after the output of the winner (“U”) is stored,
Tournament ζ: Participate. At this time, the new input record (I T I'') is the previously output record (''
It is compared with the current string (times) (2), and if the current string can be continued, the current string is distinguished as a record that belongs to C:, and if it cannot be continued, it is distinguished as the same record. Like the country of C
On the tournament tree, only comparisons are made between records belonging to the current string and between records belonging to the next string (:).If each node If a record belonging to the string f2 of the string f2 collides with a record belonging to the next string, the record belonging to the current string f2 will advance through the tournament as the winner. The last record ('"X") of the current string will be output. At the point in time when the current string has converged, the string concave rfflW"fm' is stored in the string storage means B.The string storage means 3 is composed of an external storage device E such as a magnetic disk. The next string Q of the tournament tree begins to be output as the current string ((5) to (8) in Figure 2), and the next string lff1=-. begins to be formed on the tournament ((9) in the same figure). ~(11)).

The string length monitoring means 2 detects the length of the string output to the string storage means 6 and compares it with a reference value. F exchange? The length of the string produced by F: in P part 1a is tournament (=If the number of records that can be stored is G, then the input data is completely random, which is Yonhap 2G, so the standard value is G<standard value<2G. During the period when the string length is less than the reference value, the current string generation order is determined as the input data arrangement (2) is determined to be inappropriate and is notified by the comparison unit 1bC of the string generation means 1. , the comparison unit 1b of the string generation means 1 reverses the comparison of the next string of the tournament and creates a string in the reverse order.Specifically, at the time of i21:':I (5), such an operation is performed. The generation order of the next string is changed by C2.This action changes the generation order of the strings while corresponding to the partial arrangement C2 of the input data.

The string merging unit 4 is composed of a string merging unit i4a and a string output unit 4b. When inputting all input data is completed, the string merging unit 4a+2 is notified. The string merging section 4a is the string storage means 3
Read the string from and open the merge. When merging, if the number of strings is generated in a different order, merging is performed from the beginning or the end of each string. The merged strings are returned to the string storage means 6 as new strings, and the merge is repeated. When the number of strings becomes one after the final merging, the string output section 4b outputs the result of the court noble, and the classification process ends.

Figure 2 shows an example of the generated string and merging process, and Figure 3 (2) shows the conventional method. In comparison, the conventional method requires intermediate merging processing of ≦2, whereas in this embodiment, intermediate merging can be bypassed and classification results can be output in a short time.

〔Effect of the invention〕

As explained above, the present invention dynamically changes the order of created strings while keeping an eye on the string length. This has the effect of shortening the time.

[Brief explanation of the drawing]

FIG. 1 shows the application of the present invention: 1-block area,
Fig. 2 is a diagram showing the transition state of the tournament tree in the permutation remote selection part 'Iat2 of Fig. 1, and Fig. 6 (1) and ■ are examples of strings generated by the embodiment of Fig. 1 and the conventional method, respectively. Figure i4, which shows the merging process 43 and their merging process, is a diagram of the separation process in the conventional method. 1... String generation means, 1a... Cap A selection section,
1b... Comparison section, 2... String length monitoring means,
3... String storage means, 4... String merging means, 4a... String 9 to sum iWs 4b... String output section.

Claims (1)

[Claims] A classification processing method performed by generating and merging strings, comprising: means for generating a string; means for monitoring the length of the generated string; and determining the length of the string by the monitoring. An order monitoring classification processing method comprising: means for changing the generation order of the next string when the length of the string is less than a predetermined length; and means for merging the generated strings.