JP4891979B2 - Data stream management system, record processing method, program, and recording medium - Google Patents

Description

  The present invention relates to a data stream management system that performs filter processing on records having various data types or an event processing system that performs event processing according to data, and in particular, various filter processing can be performed easily and efficiently. It relates to the technology to realize.

  First, an example of processing in a conventional data stream management system incorporating various filter processes will be described with reference to FIGS. FIG. 1 is a flowchart showing the flow of processing in the data stream management system, and FIG. 2 shows how the input record changes depending on each filter processing in a plurality of filter processing (filter processing series). FIG.

  As shown in FIG. 1, in this data stream management system, a certain record is input (step 1), and each filtering process is executed on the record until the filtering process sequence ends (steps 2 and 3). A record that has been subjected to the above process is output (step 4).

  A more specific processing example will be described with reference to FIG. In FIG. 2, the right part shows the contents of the filtering process and the record, and the left part describes what the corresponding part described on the right side is. For example, “(in11, in12, in13) → (out11, out12)” on the right side indicates that “filter processing 1” is on the left side. Hereinafter, the same applies to the same kind of figures described in this specification.

  In the example illustrated in FIG. 2, a sequence including filter processing 1, filter processing 2, and filter processing 3 as illustrated in FIG. 2 is used as the filter processing sequence.

  As shown in FIG. 2, it is assumed that (Ca, Cb, Cc) is input as an input record to the data stream management system. Then, the filter process 1 receives (Ca, Cb, Cc) as an input record and returns (Cd, Ce) as an output record. This is called current record 1. The current record 1 becomes an input record for the next filtering process 2.

  Filter process 2 receives (Cd, Ce) as an input record and returns (Cf, Cg, Ch) as an output record. This is called current record 2. The current record 2 becomes an input record for the next filtering process 3.

  Filter process 3 receives (Cf, Cg, Ch) as an input record and returns (Ci, Cj, Ck) as an output record. This is called current record 3. Since the processing sequence is completed by the end of the filter processing 3, the current record 3 becomes an output record of the data stream management system as it is.

  The filter processing method described above is a method premised on creation of the filter operation described in Non-Patent Document 1. This method is an example of managing the current record space in an update type.

  The above system has the advantage that the connection between processes is efficient and efficient. However, there is a disadvantage that the schema (data type) of the output record of the Nth filter process must match the schema of the input record of the N + 1th filter process. In other words, each filter processing in the filter processing sequence must be designed in consideration of the schema in the entire processing flow, and there is a problem that the design is difficult and dedicated to a specific purpose.

  In the above system, for example, when Ca is to be left in the final record, it is necessary to modify the filter processes 1 to 3 so that Ca is included in the output record of each filter process. The processing related to this modification is not essential logic for each filter processing, and it is desirable that the processing is originally separated.

  Next, an example in which the drawbacks of the above-described method for managing the current record space in an update type is solved will be described with reference to FIGS. The method described below is an example of a method for managing the current record space in an additional manner.

  FIG. 3 is a flowchart showing the flow of processing in the data stream management system in this example. FIG. 4 is a flowchart showing how the input record is processed by each filter process in a plurality of filter processes (filter process sequences) in this example. It is a figure which shows how it changes.

  The filter processing sequence in this example is the same as that shown in FIG. 2, and includes filter processing 1, filter processing 2, and filter processing 3. The input record and output record for each filter process are the same as those shown in FIG.

  As shown in FIG. 3, when a certain record is input (step 11), each filtering process on the record is executed until the filtering process sequence is completed (steps 11 and 14). However, in this method, the current record after the filter process is not limited to only the result record of the filter process, but is obtained by adding the process result record of the filter process to the input record to the filter process. The recorded record is made the current record (step 15 processing). In addition, an input column selection process is performed in order to construct an input record for the next filter process from the current record (the process of step 13). Further, after all the filter processes are completed, a process for selecting an output column is performed (step 16), and a record including elements of the selected column is output (step 17).

  A more specific processing example will be described with reference to FIG. As shown in FIG. 4, it is assumed that (Ca, Cb, Cc) is input as an input record to the data stream management system. Then, in the filter process 1, (Ca, Cb, Cc) is input as an input record, and (Cd, Ce) is returned as an output record. Then, addition processing 1 to the current record is performed, and (Ca, Cb, Cc, Cd, Ce) is obtained as the current record 1.

  Subsequently, the input column selection process 2 is performed immediately before the filter process 2, whereby (Cd, Ce) is selected from the current record 1 as an input record to the filter process 2. The record is filtered and (Cf, Cg, Ch) is returned. Then, addition processing 2 to the current record is performed, and (Ca, Cb, Cc, Cd, Ce, Cf, Cg, Ch) is obtained as the current record 2.

  Subsequently, the input column selection process 3 is performed immediately before the filter process 3, whereby (Cf, Cg, Ch) is selected from the current record 2 as the input record to the filter process 3. The input record is filtered and (Ci, Cj, Ck) is returned. Then, addition processing 3 to the current record is performed, and (Ca, Cb, Cc, Cd, Ce, Cf, Cg, Ch, Ci, Cj, Ck) is obtained as the current record 3.

  Finally, in the process of selecting an output column, (Ca, Ci, Cj, Ck) is selected from the current record, and the data stream management system outputs (Ca, Ci, Cj, Ck).

  The above-described method of managing the current record space in an additional type has an advantage that intermediate data other than the output record of the last filter processing such as Ca can be referred to.

  However, this method has a drawback that a shortage of resources (memory) is likely to occur due to an increase in the management area (memory area) of the current record. In particular, when the filter processing sequence is long and when media data such as images and videos is included as data, the area for storing the data tends to increase, and the system resources are compressed.

In this method, data increases each time filtering is performed, but since it is not possible to fix which of them will be referred to later, all data must be named, and name management becomes complicated (names There is also a problem that a burden when assigning and a burden of selecting a name when referring to a value are generated.
“Design and Evaluation of Additional Write / Reference Data Management System”, Akama et al., IPSJ Transactions Vol.49 No.2, February 2008, pp749-764

  As described above, in the method according to the prior art, each filter process has to be designed in consideration of the schema in the entire process flow, is difficult to design, and is dedicated. There is a problem that it is difficult to prepare a general-purpose filter process in advance.

  There is also a problem that a shortage of resources (memory) is likely to occur due to an increase in the management area of the current record. In particular, when the filter processing sequence is long, and when media data such as images and videos is included as data, the area for storing them tends to increase, and the system resources are compressed. was there.

  Furthermore, in one method of the prior art, data increases each time filtering is performed, but since it is not possible to fix which of them will be referred to later, it is necessary to name all of the data, which complicates name management There is also a problem that (the burden of assigning a name, the burden of selecting a name when referring to a value) occurs.

  The present invention has been made in view of the above points, and provides a technique for solving the above problems and realizing various filter processes easily and efficiently for records having various data types. With the goal.

In order to solve the above-described problem, the present invention provides a data stream management system that sequentially executes each filtering process in a predetermined filtering process sequence for records input as data constituting a data stream, Current record storage means for storing records, and the current record is read from the current record storage means, and the filter is applied to a local input record comprising all or part of a plurality of elements constituting the current record Filter processing execution means for executing processing, elements constituting the processing result record of the filter processing obtained by the filter processing execution means, and elements in a predetermined column from elements constituting the current record Extract and order the extracted elements in a predetermined order Created, the new current record, have a, a current record structure means for storing the current record storing means in place of the current record, the filtering process executing means a new current record by arranging, the The current record is read from the current record storage means, and the elements existing in a predetermined column are extracted from the current record so as to become an input record suitable for the filter processing, and the extracted elements are determined in a predetermined order. and the local input record structure means for creating the local input records by arranging the configured as a data stream management system, characterized by chromatic and a filter processing means for executing the filtering process.

  The data stream management system further includes a filter process execution control unit that repeatedly executes the process by the filter process execution unit and the current record forming unit in the order of filter processes in the filter process series until the filter process is completed. It is good also as having.

  According to the present invention, only necessary data can be managed thereafter in the current record, and the management load on the data storage area and the name space can be reduced. That is, when the filter processing sequence is long or when media data such as image data or video data is included as data, processing can be performed without unnecessarily increasing the area. In addition, since the management target is the minimum necessary, the management load on the name space is reduced.

  Further, according to the present invention, general-purpose filter processing can be prepared in advance and combined later. That is, the input of each filter process can be made independent of the immediately preceding current schema, and the output of each filter process can be made independent of the immediately following current schema. The relationship between the input / output of each filter process and the current schema can be flexibly adjusted. Thus, general-purpose filter processing can be applied to processing of various schemas.

  That is, according to the present invention, in the filter processing sequence for inputting multi-values and outputting multi-values, the current record composing means selects and has as much data as is really necessary as the current record. It is possible to improve the efficiency of the data storage area and prevent the name space from being confused.

  In addition, the schema of the record expected by the filtering process (applicable to the filtering process) and the schema of the current record may be different depending on the processing of the current record forming unit and the local input record forming unit. It is possible to prepare in advance independently of the record schema, that is, to prepare for general use.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments of the present invention will be described with reference to the drawings. First, terms used in the present specification will be organized and described before describing the embodiments.

(Explanation of terms)
Record: It has the same meaning as a record in a general language (C language or SQL language), and has a data type composed of a plurality of values. A record may be composed of one value. In general, the input data of the data stream management system in the present embodiment is a continuous record.

  Schema: Equivalent to the data type of the record. In this specification, the data type of the record is called a schema in accordance with the terminology of the database. As for the data type, a system specialized in text may fix the data type to one type, but it is necessary to be able to use various data types when handling multimedia data.

  Filtering process: A generic term for various processes executed on a record, and is a process performed on a record input to the filter processing unit. Filter processing includes not only processing for selecting data according to conditions but also various processing such as data integration, structuring, and replication. Note that the input record to the filter processing unit and the output record from the filter process in the present embodiment are multivalued, and the schemas of the input record and the output record may be different.

  Filter processing sequence: A sequence of a plurality of filter processes, and one record is sequentially processed according to the order of the filter processing in the sequence. Note that the filter processing sequence may include a sequence control structure such as conditional branching and repetition.

  Current record: A record stored and managed by the data stream management system as a current record (generally a record to be operated). The current record may be a record itself input from an external system or information source to the data stream management system, or may be a record reflecting the filter processing result. The content changes each time filtering is applied to the current record.

  Current schema: The schema of the current record.

  Local input record: A record to be input for filtering. Since this record is effective only for the immediately following filter processing and disappears before the next filter processing, it is called a “local” input record. The current record may become a local input record as it is.

  Local output record: An output record as a result of the previous filtering process. This record is called a “local” output record because it disappears before the next filtering process. The local output record may become the current record as it is.

  Multi-value: Multi-value means that it is not limited to only one value. Note that a general programming language function allows multi-values (a plurality of parameters) as input, but does not allow multi-values as output.

<First Embodiment>
(System configuration)
FIG. 5 is a functional configuration diagram of the data stream management system 100 according to the first embodiment of the present invention. As shown in FIG. 5, the data stream management system 100 according to the present embodiment includes a stream reception unit 101, a stream buffer unit 102, a current record management unit 103, a filter processing execution control unit 104, a result output unit 105, and a filter. It has processing parts 1-P.
For example, the stream reception unit 101 receives a stream (continuous record) via a communication network, stores the received stream data in the stream buffer unit 102, and receives an instruction from the filter processing execution control unit 104. This is a functional unit that sequentially reads each record from the stream data stored in the stream buffer unit 102 and passes it to the filter processing execution control unit 104.

  The stream buffer unit 102 is a functional unit that stores stream data, and is realized by, for example, a storage device such as a memory. The current record management unit 103 is a functional unit that stores a current record, and is realized by a storage device such as a memory.

  The filter process execution control unit 104 is a functional unit that controls the execution of each filter process in the filter process sequence targeted for the process. For each filter process, the local input record configuration process is performed on the current record to perform local input. Generate a record, input the local input record to the corresponding filter processing unit, let the filter processing unit perform the filtering process, and apply the current record composition processing to the output from the filter processing unit to generate the current record It is a functional unit that performs processing such as. That is, the filter processing execution control unit 104 includes a local input record configuration unit and a current record configuration unit.

  Further, the local input record forming unit and the filter processing unit may be collectively referred to as a filter processing executing unit. Further, the filter process execution control unit 104 includes a filter process execution control unit that repeatedly executes the process by the filter process execution unit and the current record forming unit in the order of the filter processes in the filter process series until the filter process is completed. Have.

  Details of the local input record composition process and the current record composition process will be described later.

  Each of the filter processing units 1 to P is a functional unit that executes filter processing on the local input record and outputs the result to the filter processing execution control unit 104. The result output unit 105 is a functional unit that receives the current record after all the filter processes included in the filter process sequence are executed from the filter process execution control unit 104, and outputs the current record to the outside via a communication network, for example.

  The data stream management system 100 including the above functional units can be realized, for example, by causing a computer to execute a program corresponding to the processing of each functional unit. The program may be stored in a recording medium such as a memory or a CD-ROM, installed on a computer from the recording medium and executed, or downloaded and executed by a computer from a server on a network. Also good.

(System operation)
Next, a basic operation of the data stream management system 100 in the present embodiment will be described with reference to FIGS.

  First, the overall processing flow will be described with reference to FIG. FIG. 6 shows processing related to one record among records continuously input to the data stream management system 100. Actually, the process shown in FIG. 6 is continuously performed for each input record. Further, as a premise of the processing in FIG. 6, it is assumed that the current record to be processed (the record that has been input at the beginning of the processing) is stored in the current record management unit 103. FIG. 7 is a diagram showing a specific example of a record and showing the processing contents of FIG. 6 in an easy-to-understand manner. In the description of the processing of FIG. 6 below, the contents of the record in FIG.

  As shown in FIG. 6, the filter processing execution control unit 104 checks whether or not all the filter processing in the filter processing sequence has been completed (step 101).

If all the filter processes in the filter process sequence have not been completed, a process related to a filter process (referred to as filter process f _N ) next to the filter process that was the previous process target is performed. Also, the current record at this point is defined as current record N.

In step 102 of FIG. 6, the filter process execution control unit 104 acquires the current record N from the current record management unit 103, and performs local input record N configuration processing on the current record N (step 103). That is, carried out from the current record N, the process of creating a local input record N as an input record for filter processing unit that performs a filtering process f _N. In the example of FIG. 7, a local input record N (Ca2, Ca3) is created from the current record N (Ca1, Ca2, Ca3).

Subsequently, the filter processing execution control unit 104 receives the local input records N in the filter processing unit that performs filter processing f _N, the filter processing unit performs filter processing f _N to the local input record N Then, the local output record N that is the result of the filtering process f _N is output to the filtering process execution control unit 104 (step 104 in FIG. 6). In the example of FIG. 7, filtering f _N is subjected to a local output records N (Cb1, Cb2) are obtained for (Ca2, Ca3).

  Subsequently, the filter process execution control unit 104 performs a current record N + 1 configuration process on the local output record N (step 105 in FIG. 6). That is, the current record N + 1 that is the next current record of the current record N is created from the local output record N and the like, the created current record N + 1 is stored in the current record management unit 103, and the immediately preceding current record Process to discard N. In the example of FIG. 7, the current record N + 1 (Cb1, Ca3, Cb2, Ca1) is obtained from the current record N (Ca1, Ca2, Ca3) and the local output record N (Cb1, Cb2).

  Thereafter, the process from step 101 shown in FIG. 6 is executed again with N + 1 as a new N.

  In step 101, when all the filter processes in the filter process sequence have been completed, the filter process execution control unit 104 uses the current record N stored in the current record management unit 103 as the process result of the filter process sequence. The result is output to the result output unit 105, and the result output unit 105 outputs the record (step 106).

Next, the current record N + 1 configuration process will be described with reference to FIG. In the current record N + 1 configuration process, the filter process execution control unit 104 calculates the current from the local output record N that is the output result record of the filter process f _N and the current record N that is the current record after the previous filter process. A predetermined partial column set is selected as a set of columns in the record N + 1 (step 201). Note that a column is information indicating the location of an element in a record, but may also be used to mean the element at that location. Subsequently, the filter processing execution control unit 104 creates a current record candidate by arranging the selected column set in a predetermined order (step 202). Then, the filter process execution control unit 104 discards the current record N from the current record management unit 103, and stores the current record candidate created in step 202 as the current record N + 1 in the current record management unit 103 (step 203).

Next, the local input record N configuration process will be described with reference to FIG. In the local input record N configuration process, the filter process execution control unit 104 selects a partial set of columns determined in advance as being suitable for the filter process f _N from the current record N (step 301). Then, the filter processing execution control unit 104, a set of the column to form a local input record N by arranging in a predetermined order conforming to filter f _N (step 302).

  Next, the operation of the data stream management system 100 according to the present embodiment will be described based on a more specific example with reference to FIG.

  The filter processing sequence in this example is the same as that shown in FIG. 2, and includes filter processing 1, filter processing 2, and filter processing 3. The input record and output record for each filter process are the same as those shown in FIG. In the following description, it is assumed that filter processing 1, filter processing 2, and filter processing 3 execute filter processing 1, filter processing 2, and filter processing 3, respectively.

  As shown in FIG. 10, (Ca, Cb, Cc) is input as an input record to the data stream management system 100, and (Ca, Cb, Cc) is stored in the current record management unit 103 as current record 1. . Next, the local input record 1 configuration process is performed. Here, since the current record 1 itself becomes the local input record 1, no special process is performed as the local input record 1 configuration process.

  Subsequently, the filter processing execution control unit 104 inputs the local input record 1 (Ca, Cb, Cc) to the filter processing unit 1, and the filter processing unit 1 applies to the local input record 1 (Ca, Cb, Cc). The filter process 1 is executed and the local output record 1 (Cd, Ce) is output to the filter process execution control unit 104. Then, the filter process execution control unit 104 performs the current record 2 configuration process, so that the current record 2 (Ca, Cd) is obtained from the current record 1 (Ca, Cb, Cc) and the local output record 1 (Cd, Ce). , Ce), and stores it in the current record management unit 104 and discards the current record 1 (Ca, Cb, Cc).

  Next, the filter process execution control unit 104 acquires the local input record 2 (Cd, Ce) by executing the local input record 2 configuration process for the current record 2 (Ca, Cd, Ce), and performs local input. Record 2 (Cd, Ce) is input to the filter processing unit 2. The filter processing unit 2 executes the filter processing 2 on the local input record 2 (Cd, Ce) and outputs the local output record 2 (Cf, Cg, Ch) to the filter processing execution control unit 104. Then, the filter process execution control unit 104 performs the current record 3 configuration process to obtain the current record 3 (Ca, Cg, Ch) from the current record 2 (Ca, Cd, Ce) and the local output record 2 (Cf, Cg, Ch). , Cf, Cg, Ch), and stores it in the current record management unit 103, and discards the current record 2 (Ca, Cd, Ce).

  Next, the filter process execution control unit 104 acquires the local input record 3 (Cf, Cg, Ch) by executing the local input record 3 configuration process for the current record 3 (Ca, Cf, Cg, Ch). Then, the local input record 3 (Cf, Cg, Ch) is input to the filter processing unit 3. The filter processing unit 3 executes the filter processing 3 on the local input record 3 (Cf, Cg, Ch) and outputs the local output record 3 (Ci, Cj, Ck) to the filter processing execution control unit 104. Then, the filter process execution control unit 104 performs the current record 4 configuration process, so that the current record 4 is obtained from the current record 3 (Ca, Cf, Cg, Ch) and the local output record 3 (Ci, Cj, Ck). (Ca, Ci, Cj, Ck) is acquired, stored in the current record management unit 103, and the current record 3 (Ca, Cf, Cg, Ch) is discarded.

  In this example, since all the filter processes are completed as described above, the current record 4 (Ca, Ci, Cj, Ck) is output from the result output unit 105 as a process result by the filter process sequence.

<Second Embodiment>
Next, an embodiment using a more specific processing sequence as the filtering processing sequence will be described as a second embodiment. In the present embodiment, the functional configuration of the data stream management system 100 and the basic processing flow are the same as those described in the first embodiment.

  In the present embodiment, the following filter processing is used as the filter processing used in the filter processing sequence. In the description of the filter processing below, the description of the filter processing is described on the right side of the name of the filter processing, and the input / output is shown below the description. Each input / output is described in the order of input / output data name and type (form). For example, in the first filtering process, ppm2jpg, InImg is the name of the input data, and ppm indicates the type.

  ppm2jpg: Converts PPM format image data to JPEG format image data.

Input: InImg ppm, Output: OutImg jpg
jpg2ppm: Converts JPEG image data to PPM image data.

Input: InImg jpg, Output: OutImg ppm
findFace: The largest face area in the PGM format input image is detected, and the rectangular area information is returned to X0, X1, Y0, and Y1. X0 and X1 are X coordinate values, and Y0 and Y1 are Y coordinate values.

Input: InImg pgm, Output: X0 int, X1 int, Y0 int, Y1 int
resize: Changes the size of input image data in PPM format to vertical H horizontal W.

Input: InImg ppm, filter arguments: H int, W int, output: OutImg ppm
rgb2hsi: Converts input image data in PPM format to each grayscale image data (PGM format) in HSV space.

Input: InImg ppm, Output: HueImg pgm, SatImg pgm, IntImg pgm
drawCircle: Draws an ellipse inscribed in a rectangle (X0, Y0, X1, Y1) in image data in PPM format.

Input: InImg ppm, X0 int, Y0 int, X1 int, Y1 int, Output: OutImg ppm
writeText: Draws a text string T given by SJIS on PPM format image data.

Input: InImg ppm, T text, Output: OutImg ppm
toSJIS: Converts the input character string to SJIS format.

Input: InT text, Output: OutT text
Each filter processing unit in the data stream management system 100 according to the present embodiment has a function corresponding to the above filter processing, and the filter processing execution control unit 104 performs local input record configuration processing and current processing suitable for each filter processing. It has a function to perform record composition processing.

  In the present embodiment, it is assumed that data of the following schema continues to be given as an input to the data stream management system 100.

  IMG jpg: Video data from the camera, in MotionJPG format.

  T text: text information from the camera, for example, text information indicating the outside air temperature.

  In the present embodiment, it is assumed that the data stream management system 100 executes a filter processing sequence including the following series of filter processing.

jpg2ppm ();
resize (H = 400, W = 640);
rgb2hsi ();
findface ();
drawCircle ();
toSJIS ();
writeText ();
ppm2jpg ();
FIG. 11 shows a description example representing processing executed by the data stream management system 100 according to the present embodiment. In this example, the current record N + 1 composition process is named MRGandSEL, the local input record N composition process is named SELforLOCAL, and the result of the previous filter process (local output record N) is the name space RES. Temporarily manage, refer to the result, and give a new name with as. The processing logic of MRGandSEL and SELforLOCAL related to each filter processing is incorporated in the filter processing execution control unit 104 by a program (script).

  The basic processing flow according to the description shown in FIG. 11 is the same as that shown in FIGS. 6 to 9 of the first embodiment.

  For example, as the processing related to the filter processing resize in the second row, first, the filter processing execution control unit 104 acquires the current record (referred to as current record 2) after the filter processing in the first row from the current record management unit 103. . This current record 2 is generated by the current record 2 configuration process MRGandSEL described in the first line, and is specifically (T, IMGppm). Subsequently, the filter processing execution control unit 104 extracts IMGppm from (T, IMGppm) as the local input record 2 configuration process SELforLOCAL, inputs IMGppm to the filter processing unit that performs the filter processing resize, and performs the filter processing resize. The current record 3 configuration process MRGandSEL is performed with OutImg being the process result as IMGsz, and current record 3 (T, IMGsz) is acquired. The same processing is performed with respect to other filter processing except for the difference in processing contents.

  When trying to execute such a filter processing sequence, the conventional method of managing the current record space with the update type cannot be executed in the first place, and the conventional method of managing the current record space with the additional type cannot be executed in the first place. A wasteful ppm image is generated.

  On the other hand, according to the method shown in the present embodiment, the current record N + 1 configuration processing (MRGandSEL) and the local input record N configuration processing (SELforLOCAL) are applied to an existing filter processing sequence having no consistency of input / output schema. By combining these, it becomes possible to configure an existing filter processing sequence as a series of filter processing sequences. Furthermore, according to the method according to the present embodiment, it is sufficient to hold data as much as necessary for subsequent processing as a current stream, and resources are not wasted.

  The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

It is a flowchart which shows the flow of a process in the conventional data stream management system. It is a figure which shows the change of the record accompanying a filter process in the conventional data stream management system. It is a flowchart which shows the flow of a process in the conventional data stream management system. It is a figure which shows the change of the record accompanying a filter process in the conventional data stream management system. It is a functional lineblock diagram of data stream management system 100 concerning a 1st embodiment of the present invention. It is a flowchart which shows operation | movement of the data stream management system 100 which concerns on 1st Embodiment. It is a figure for demonstrating the processing content of FIG. 6 concretely. It is a flowchart which shows a current record N + 1 composition process. It is a flowchart which shows a local input record N structure process. It is a figure for demonstrating operation | movement of the data stream management system 100 which concerns on 1st Embodiment based on a specific example. It is a description example showing the process which the data stream management system 100 which concerns on 2nd Embodiment performs.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Data stream management system 101 Stream reception part 102 Stream buffer part 103 Current record management part 104 Filter process execution control part 105 Result output part 1-P Filter process part

Claims (6)

A data stream management system that sequentially executes each filter processing in a predetermined filter processing sequence for records input as data constituting a data stream,
Current record storage means for storing the current record;
Filter processing execution means for reading out the current record from the current record storage means, and executing the filter processing on a local input record consisting of all or part of a plurality of elements constituting the current record;
The elements in the predetermined column are extracted from the elements constituting the processing result record of the filtering process obtained by the filtering process execution unit and the elements constituting the current record, and the extracted elements are determined in advance. create a new current record by arranging the order was, the new current record, have a, a current record structure means for storing the current record storing means in place of the current record,
The filtering process execution means
The current record is read from the current record storage means, and an element existing in a predetermined column is extracted from the current record so as to become an input record suitable for the filtering process, and the extracted element is determined in advance. Local input record constructing means for creating the local input record by arranging in order;
Data stream management system, characterized by chromatic and a filter processing means for executing the filtering process. Claim 1, wherein the filtering process execution means, and processing by the current record configuration means, in the order of the filtering process in the filtering process sequence a filter process execution control means for repeatedly executed until the filter processing is completed The data stream management system described in 1. A record processing method executed in a data stream management system that sequentially executes each filter processing in a predetermined filter processing sequence for records input as data constituting a data stream,
The data stream management system includes current record storage means for storing a current record, and the record processing method includes:
The filter processing execution means provided in the data stream management system reads the current record from the current record storage means, and for a local input record consisting of all or a part of a plurality of elements constituting the current record A filtering process executing step for executing the filtering process;
The current record constituting means provided in the data stream management system is arranged in a predetermined column from elements constituting the processing result record of the filtering process obtained by the filtering process executing step and elements constituting the current record. A current record that extracts a certain element, creates a new current record by arranging the extracted elements in a predetermined order, and stores the new current record in the current record storage means instead of the current record configuration and step, the possess,
The filtering process execution step includes:
The current record is read from the current record storage means, and an element existing in a predetermined column is extracted from the current record so as to become an input record suitable for the filtering process, and the extracted element is determined in advance. A local input record construction step for creating the local input record by arranging in order;
Record processing method characterized by chromatic and a filtering processing step of executing the filtering process. The filter processing execution control means included in the data stream management system repeatedly executes the processing by the filtering processing execution step and the current record composition step in the order of filtering processing in the filtering processing series until the filtering processing is completed. The record processing method according to claim 3 . The program for functioning a computer as each means with which the data stream management system of Claim 1 or 2 is provided. A computer-readable recording medium on which the program according to claim 5 is recorded.

Images