JP2015028701A - Sequence extraction method, sequence extraction program, and sequence extraction device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method of extracting and illustrating an operation sequence in a document.SOLUTION: A sequence extraction method includes: a structure data generation step of segmenting sentences constituting a document into paragraphs based upon time condition words and punctuation marks, and generating structure data including each paragraph, a first identifier indicative of the paragraph, a second identifier indicative of a sentence including the paragraph, and a time condition word; a time condition extraction step of extracting the time condition word from the structure data, and generating time condition data including the time condition word, a transition direction corresponding to the time condition word, and the first identifier corresponding to the time condition word; and a sequence extraction step of extracting the time condition word from the time condition data, extracting a first identifier corresponding to a paragraph having the same identifier as the second identifier corresponding to the time condition word from the structure data, and generating operation sequence data including the first identifier corresponding to the time condition word, a first identifier of a paragraph having the second identifier, and a time direction corresponding to the time condition word for the second identifier.

Description

  The present invention relates to a method, a program, and an apparatus for sequence extraction.

  In the development and design of an LSI (Large Scale Integration), a designer designs the LSI in accordance with IC (Integrated Circuit) standards and specifications. The verifier verifies whether the design of the designer is designed according to the standard or specification. In recent years, with the advancement of technology development, IC standards and specifications have become more complex with the increasing number of LSI functions. On the other hand, the period from development to manufacture of LSI is shortened. The designer reads a standard document or a specification document in which a huge amount of standards or specifications are written in a short period of time and designs to satisfy the standard or specification. The verifier constructs a verification environment and a test pattern in order to verify whether the standard or specification is satisfied in a short period of time. However, due to the complexity of the standards and specifications described above, it is difficult for the designer to analyze the standards and specifications, extract the functions, and perform the design after fully understanding. In addition, it is very difficult for a verifier to construct a verification environment and a test pattern for verifying whether a designer's design satisfies standards and specifications.

  Standards and specifications that are the basis of design are generally written in natural language. A sentence written in a natural language expresses an operation sequence such as an operation condition or order in the LSI in characters. Designers and verifiers must grasp the conditions and order of operations from a large amount of text, and oversight often occurs. Therefore, it is desired to exhaustively extract operation conditions and order.

  Natural language analysis is applied to extract information such as operation conditions and order from sentences written in natural language. For example, Patent Document 1 discloses a technique for automatically interpreting the semantic correspondence between a plurality of input sentences by natural language analysis. Patent Document 1 discloses a technique for analyzing a time series relationship by interpreting a tense of a plurality of sentences. Furthermore, Non-Patent Document 1 discloses a technique for extracting a keyword from a sentence such as a Web page written in a natural language and extracting the relationship between the sentence and a time zone.

JP 2006-190101 A

Taichi Noro, 3 others, “Determining the occurrence time zone of events in text”, Journal of Information Processing Society of Japan, October 2007, Vol. 48 No. 10, p. 3405-3414

  However, with the above-described conventional technology, even if a time series of events in a sentence can be extracted, it is difficult to extract an operation sequence or state transition of an LSI or the like represented in a standard or specification.

  An object of the present invention is to provide a method, a program, and an apparatus that extract an operation sequence and state transition in a sentence and illustrate them to a designer and a verifier in order to solve the above-described problems caused by the prior art.

  In order to solve the above-described problems and achieve the object, according to one aspect of the present invention, a time condition word that leads a sentence having a time or condition element to a sentence constituting the sentence with respect to the sentence to be analyzed. And a punctuation segment into sections, each section including the section, a first identifier indicating the section, a second identifier indicating a sentence including the section, and the section includes a time condition word A structure data generation step for creating structure data including the time condition word, extracting the time condition word from the structure data, the time condition word, and a transition direction corresponding to the time condition word, A time condition extracting step for creating time condition data including a first identifier corresponding to the time condition word; extracting a time condition word from the time condition data; and corresponding to the time condition word from the structure data The same as the second identifier A first identifier corresponding to a clause having two identifiers, and for the second identifier, the first identifier corresponding to the time condition word and the first identifier of the clause having the second identifier A sequence extraction method, a program, and an apparatus are proposed that include a sequence extraction step of creating operation sequence data including an identifier and a time direction corresponding to the time condition word.

  According to one aspect of the present invention, an operation sequence or state transition in a sentence can be extracted and shown to a designer or a verifier. By referring to the illustrated operation sequence and state transition, the designer can design with a complete understanding of the standards and specifications. In addition, by referring to the illustrated operation sequence and state transition, the verifier can easily construct a verification environment and a test pattern for verifying whether the design of the designer satisfies the standard or specification. The effect of becoming.

FIG. 1 is an explanatory diagram showing the configuration of the order extraction apparatus according to the present embodiment. FIG. 2 is an output image diagram of an existing structural analysis tool. FIG. 3 is an image diagram of input / output according to the present embodiment. FIG. 4 is a block diagram illustrating an example of a hardware configuration of the order extraction device. FIG. 5 is an overall flowchart from reading a file to generating structure data. FIG. 6 is a flowchart for generating the structure data. FIG. 7 is a flowchart for dividing a sentence into clauses in order to generate structure data. FIG. 8 is a flowchart for dividing a sentence into clauses in order to generate structure data. FIG. 9 is a flowchart for extracting a sequence. FIG. 10 is a flowchart for extracting state transitions. FIG. 11 is a flowchart for extracting related clauses. FIG. 12 shows TC dictionary data representing the time condition direction of a specific word. FIG. 13 is an example showing the contents of the sentence DB 111. FIG. 14 is an example showing the contents of the state transition data. FIG. 14 is an image diagram in which state transitions are extracted and illustrated to the verifier.

  DESCRIPTION OF EMBODIMENTS Embodiments of a disclosed order extraction apparatus, method, and program will be described in detail below with reference to the accompanying drawings.

FIG. 1 is an explanatory diagram showing the configuration of the order extraction apparatus according to the present embodiment. The order extraction apparatus 100 according to the present embodiment is an apparatus that extracts and visualizes an order from a sentence. The order extraction apparatus 100 includes a structure data generation unit 101, a time / information analysis unit 102, a time condition extraction unit 102, and a state transition extraction unit 104, and includes a TC dictionary 110, a sentence DB 111, and state transition ID correspondence data 112 as a database. . Standards and specifications written in a natural language to be extracted are stored in the sentence DB 120.
The TC dictionary 110 is a database in which time condition words (TC words) and time condition directions are combined. A time condition word is a word having elements of time (Time) and condition (Condition) such as “After” and “When”. In terms of grammatical terms, the time condition word is “a conjunction that leads to the adverbial clause of time”. When applied to an example of a sentence whose analysis target is a standard or specification, the time condition word is a conjunction that leads to an adverbial clause describing the execution timing and execution condition of the internal operation of the LSI.

  FIG. 2 is an output image diagram obtained by analyzing a sentence written in a specification by an existing structural analysis tool. As the text written in the specification, for example, the text of Formula 1 is input.

An existing structural analysis tool analyzes the structure of Equation 1 to obtain FIG. In FIG. 2, S (sentence) 201 and 203 indicate nodes, and PN (puncture) 202 indicates a reading point. Equation 2 is divided into two clauses at the reading point 202. The existing structure analysis tool analyzes the structure of the sections 201 and 202 up to the word level.
FIG. 3 is an image diagram in which the order extracting device analyzes the text written in the specification and extracts the order. The operation of the order extraction device for the sentence of Equation 1 will be described. In FIG. 3, “Configure Endpoint Command” is displayed as “CEC” and “SET_CONFIGURATION” is displayed as “SC” in order to improve the visibility of the illustration.

  The structure data generation unit 101 performs morphological analysis and dependency analysis of the sentence of Equation 1. Since morphological analysis and dependency analysis are existing technologies, details are omitted. As a result of dependency analysis, it is found that the word “request” depends on CEC 301 and SC 302. Group 304 summarizes the state transition requirements for CEC 301 and SC 302.

  The group 307 is a transition destination when the condition of the group 304 is satisfied. As the time condition word “After” 303 of the group 304 indicates, when the group 304 is completed, the process proceeds to the group 307. The group 307 stores the software and TDs related to the software 305 that is the operation subject and the TDs 306 that is the operation target.

  FIG. 4 is a block diagram illustrating an example of a hardware configuration of the order extraction device 100. In FIG. 4, the order extraction device 100 includes a central processing unit (CPU) 401, a read-only memory (ROM) 402, and a random access memory (RAM) 403. In addition, the order extraction device 100 includes a storage device 404 and an interface 408. The order extraction device 100 includes a display 305, a keyboard 406, and a mouse 407. Further, the CPU 401 to the interface 408 are connected to each other by a bus 400.

  The CPU 401 is an arithmetic processing device that controls the entire order extracting device 400. The ROM 402 is a non-volatile memory that stores a program such as a boot program. A RAM 403 is a volatile memory used as a work area for the CPU 401.

  The storage device 404 includes an internal storage device 409 that holds the storage medium inside and an external storage device 410 that can take out the storage medium to the outside. The storage device 404 is a control device that controls reading and writing of data with respect to the internal storage medium 411 or the external storage medium 412 according to the control of the CPU 401. A magnetic disk drive or a solid state drive (SSD) can be selected as the internal storage device 409 according to the type of the internal storage medium 411. The external storage device 410 is, for example, a magnetic tape device, a magnetic disk drive, a magneto-optical disk drive, an optical disk drive, or SDD. The external storage medium 412 is a non-volatile memory that stores data written under the control of the external storage device 410. For example, when the external storage device 410 is a magnetic disk drive, the external storage medium 412 is a flexible disk (FD) that is a magnetic disk, for example. When the external storage device 410 is an optical disk drive, the external storage medium 412 is, for example, a Compact Disc Recordable (CD-R) that is an optical disk. When the external storage device is an SDD, the external storage medium is a semiconductor element memory.

  The display 405 is a device that displays data such as a cursor, an icon, or a tool box, as well as documents, images, function information, and the like. The display 405 is, for example, a Cathode Ray Tube (CRT), a Thin Film Transistor (TFT) liquid crystal display, a plasma display, an organic electroluminescence display, or the like.

  The keyboard 406 has keys for inputting characters, numbers, various instructions, and the like, and is a device for inputting data. The keyboard 406 may be a touch panel type input pad or a numeric keypad. The mouse 407 is a device for moving a cursor, selecting a range, moving a window, changing a size, and the like. The mouse 407 may be a trackball or a joystick as long as it has the same function as a pointing device.

  The interface 408 is a control device that is connected to the network 220 and controls input / output of data from an external device. The interface 408 corresponds to the type of the network 420 to be connected. The interface 408 is, for example, a wired interface or a wireless LAN Network Interface Card (NIC).

  FIG. 5 is an overall flow of the present invention. The structure data generation unit 101 reads a file stored in the sentence DB (step S501). The structure data generation unit 101 acquires the number of sentences out of sentences in the file (step S502). The structure data generation unit 101 performs the processing of steps S504 to S507 for each of the first sentence to the last sentence of the sentences in the file (steps S503 to S508). The structure data generation unit 101 assigns a unique identifier (sentence ID) to the sentence (step S504). Next, the structural data generation unit 101 performs morphological analysis (step S505) and structural analysis (step S506) on the sentence. Here, morphological analysis (step S505) and structural analysis (step S506) can use existing natural language analysis techniques. Subsequently, the structure data generation unit 101 generates structure data (step S507). The structure data will be described later. The structure data generation unit 101 performs steps S504 to S507 until the last sentence is completed (step S508).

FIG. 6 is a flow of structure data generation (step S507). First, the structure data generation unit 101 determines whether a comma is included in a sentence (step S601). When the sentence includes a comma, the structure data generation unit 101 extracts a sentence from the head of the sentence to the part before the comma as a first section and a part from the comma to the part before the period as a second section (step S602). Next, the structure data generation unit 101 performs the processing of step S604 and step S605 on both the first clause and the second clause (steps S603 to S606). The structure data generation unit 101 assigns a unique identifier (section ID) to the section (step S604). The section ID can be given the sentence ID as a prefix and the order of the sections as a child number. For example, when the sentence ID is “AAA”, the section ID of the first section can be “AAA.1” and the section ID of the second section can be “AAA.2”. Thereafter, the structure data generation unit 101 performs node subdivision 1 for subdividing the nodes (step S605). The structure data generation unit 101 performs the processes in steps S604 and S605 until the last section ends (step S606).

  When the structure data generation unit 101 determines that the sentence does not include a comma, the structure data generation unit 101 determines whether the sentence has a time condition word next (step S610). The structure data generation unit 101 ends when the time condition word is not included in the sentence. When the structure data generation unit 101 determines that the time condition word is included in the sentence, the structure data generation unit 101 determines whether the time condition word is at the beginning of the sentence (step S611). The structure data generation unit 101 ends when there is a time condition word at the beginning of the sentence. When there is no time condition word at the beginning of the sentence, the structure data generation unit 101 extracts the section from the beginning of the sentence to the time condition word as the third section and the section from the time condition word to the period before as the fourth section (step S612). Next, the structure data generation unit 101 performs the process of step S614 on each of the third and fourth sections (steps S613 and S615). The structure data generation unit 101 assigns a unique identifier to the clause (step S614). Such an identifier can be attached by the method described in step S604. The structure data generation unit 101 performs the process of step S614 until the last clause ends (step S615).

  FIG. 7 is a flow of the subdivision 1 described in step S605 for subdividing a sentence into sections. The structure data generation unit 101 determines whether there is a time condition word in the clause (step S701). The structure data generation unit 101 ends when the time condition word is not included in the clause. When the structure data generation unit 101 determines that the time condition word is included in the clause, the structure data generation unit 101 determines whether the time condition word is at the head of the clause (step S702). When the structure data generation unit 101 determines that there is no time condition word at the beginning of a clause, the structure data generation unit 101 divides the beginning of the clause to the front of the time condition word and the time condition word to the end of the clause as new clauses ( Step S703). When there is a time condition word at the beginning of a clause, the structure data generation unit 101 assigns a flag representing the beginning of the sentence to the clause (step S704). The structure data generation unit 101 extracts from the time condition word at the beginning of the sentence to the end of the section as a section (step S705), and stores the section ID, the time condition word, and the sentence head flag in the sentence DB 111 (step S706).

  FIG. 8 shows a specific example of the relationship between sentences and clauses after the flow of FIGS. The structure data generation unit 101 assigns a sentence ID “AAA” to the sentence 801 of Equation 1 (step S504), performs morphological analysis (step S505), and performs structure analysis (step S506), and then generates the structure data of FIG. Perform the process described in the flow.

  The structure data generation unit 101 determines whether the sentence 801 includes a comma (step S601). Here, since the expression 1 includes a comma, the structure data generation unit 101 extracts a node 802 from the head of the sentence to the front of the comma and a node 803 from the comma to the period, respectively (step S602). Next, the structure data generation unit 101 performs the following processing for both the node 802 and the node 803 (steps S603 to S606). The structure data generation unit 101 assigns a clause ID “AAA.1” prefixed with the statement ID “AAA” of the statement 801 to the clause 802 (step S604). Here, the child numbers “.1” are assigned in ascending order from the beginning of the sentence. Next, the structure data generation unit 101 tries to subdivide the node 802 (step S505).

  The structure data generation unit 101 determines whether a time condition word is included in the clause 802 (step S701). Since “After”, which is a time condition word, exists in the clause 802, the structure data generation unit 101 next determines whether there is a time condition word at the beginning of the clause 802 (step S702). Since the time condition word “After” is present at the beginning of the clause 802, the structure data generating unit 101 assigns a beginning flag to the clause 802 (step S704). Next, the structure data generation unit 101 extracts from the time condition word to the end of the clause as a clause (step S705). In other words, the sentence from “After” to the end of the sentence, that is, the section 802 itself is extracted as a section. The structure data generation unit 101 stores the section ID “AAA.1”, the time condition word “After”, and the sentence head flag in the sentence DB 111 in association with the section 802 (step S706).

  FIG. 9 is a flow for extracting a time condition. Time Condition The time condition extraction unit 102 selects one line from the sentence DB 111 (step S901). The time condition extraction unit 102 selects a time condition word from the selected line (step S902). Next, the time condition extraction unit 102 determines whether or not a sentence head flag is added in the selected line (step S903). When it is determined that the sentence head flag is not attached, the time condition extraction unit 102 searches the TC dictionary, acquires the direction of the selected time condition word (step S904), and ends. If it is determined that the sentence head flag is attached, the time condition extraction unit 102 searches the TC dictionary to obtain a direction when the selected time condition word is present at the beginning of the sentence (step S905), and selects the sentence DB 111. The direction is recorded in the line (step S906).

  The time condition word may exist at the beginning of the sentence or may exist at the beginning of the sentence. In the sentence 801 of Formula 1, a time condition word “After” is present at the beginning of the sentence. In this case, the content of the latter half 803 occurs after the content of the clause 802 including the time condition word “After”. On the other hand, the case of Formula 2 where the time condition word “After” is not at the beginning of the sentence will be described.

When the processing of FIGS. 5 to 7 is performed on the sentence of Expression 2, the sentence is divided into a clause from the beginning of the sentence to the time condition word “After” and a clause including the time condition word “After”. The content of the clause containing the time condition word “After” comes first, and the content of the clause before the time condition word “After” occurs after the beginning of the sentence. That is, even in the time condition word expressed by the same word, the time / condition direction of the sentence structure changes depending on whether it exists at the beginning of the sentence. If the sentence DB 111 has a sentence flag, the direction when the time condition word is present at the beginning of the sentence is stored in the sentence DB 111 (step S906).

  FIG. 10 is a flow for extracting state transitions. The state transition extraction unit 104 performs the following processing for all the rows in the sentence DB 111 (steps S1001 and S1006). The state transition extraction unit 104 selects one line of the sentence DB 111 and sets the clause stored in the line as the first entity (step S1002). The state transition extraction unit searches the sentence DB 111 whether there is a clause that matches the first entity (step S1003). If there is a matching clause, the state transition extraction unit 104 extracts related information (step S1004) and stores it in the state transition data (step S1005). When the processing is completed for all the rows, the state transition extraction is finished (step S1006).

  FIG. 11 is a flow for extracting related information. The state transition extraction unit 104 acquires hierarchical information based on the number of times the sentence ID and the child number are assigned from the section ID of the row to which the section (second entity) that matches the first entity described above belongs (step S1101). . Next, the state transition extraction unit determines whether the hierarchy information is 1 or more (step S1102). If the hierarchy information is 1 or more, the state transition extraction unit searches the sentence DB for a section (third entity) of a line having the same sentence ID and the same hierarchy information (S1103). Next, the state transition extraction unit determines whether the second entity and the third entity are connected by a time condition word (step S1104). Specifically, determination is made based on the “direction” belonging to the row of the second entity, and the prior relationship between the node ID of the second entity and the node ID of the third entity (step S1104). When the “direction” belonging to the row of the second entity is “right” and the clause ID of the third entity is after the clause ID of the second entity, point to the clause ID of the second entity It is determined that the connection is made first, and the third entity is acquired (step S1105). When the “direction” belonging to the row of the second entity is “left” and the node ID of the third entity is before the node ID of the second entity, the node ID of the second entity is pointed It is determined that the connection is made first, and the third entity is acquired (step S1106). Otherwise, the process ends.

  FIG. 12 shows an example of a TC dictionary 1200 which is a database in which time condition words and time condition directions are combined. For example, for each time condition word 1202, a time condition direction 1201 indicating the direction of passage of time and the relationship due to the establishment of the condition is recorded. When attention is paid to the row 1203 in which the time condition word 1202 is “After”, the time condition direction is “right”. A clause having the time condition word “After” at the beginning of the sentence indicates that there is a relationship with the right clause, that is, the clause following the clause. On the other hand, when attention is paid to the row 1204 in which the time condition word 1202 is “after”, the time condition direction is “left”. A clause with after indicates that it has a relationship with the left clause, that is, the clause before that clause.

FIG. 13 is an example showing the contents of the sentence DB 111. The sentence DB 1300 includes a section ID 1301, a time condition word 1302, a sentence head flag 1303, a section entity 1304, and a direction 1305 as constituent elements.
Focusing on the row 1306, “AAA.1” is recorded as the node ID 1301. From this, it can be seen that the section in the row 1306 is a part of the sentence whose sentence ID is “AAA”. In the time condition word 1302, “After” is recorded. Then, “right” is recorded in the direction 1305, and “1” is recorded in the sentence flag 1303. From these facts, the section in the row 1306 has a relationship with the section on the right side, that is, the section following the section having the section ID “AAA.1”, with the time condition word “After” at the beginning of the sentence. It can be seen that there is a clause. In the clause entity 1304, “After the CEC and SC request complete successfully” is recorded.

When attention is paid to the row 1307, the time condition word 1302 “after” is recorded. The direction 1305 is recorded as “left”, and the sentence head flag 1303 is recorded as “0”. From these facts, the section on line 1307 has a time condition word “after” that is not the beginning of the sentence, and the section on the left side, that is, the section “BBB.1” before the section with the section ID “BBB.2”. It can be seen that the nodes have a relationship between. In the clause entity 1304, “after the CEC request complete successfully” is recorded.

FIG. 14 is an example of state transition data. In the state transition data 1400, a section ID 1401, a section entity 1402, a point destination section ID 1403, and a time condition word 1404 are recorded. The point destination section ID 1403 indicates the transition destination to which section the current section transitions. The time condition word 1404 stores a time condition word related to the transition.
When attention is paid to the section in the row 1405, the section ID 1401 records “AAA.1”. In the clause entity 1402, “After the CEC and SC request complete successfully” is recorded. In the time condition word 1404, the time condition word “After” is recorded. The point destination section ID 1403 indicates that the section entity transitions to the section “AAA.2”.

When attention is paid to the node in the row 1406, the node ID 1401 records “BBB.2”. In the clause entity 1402, “after the CEC request complete successfully” is recorded. A time condition word “after” is recorded in the time condition word 1404. The point destination section ID 1403 indicates that the section entity transitions to the section entity “BBB.1”. The time condition word “after” affects the transition to the “BBB.1” clause entity.

  FIG. 15 is an image diagram of an execution result displayed upon implementation in the present invention. The output image 1500 is a visualization of the state transition of Equation 1. An object 1501 is an entity of the clause 802 in Equation 1. The object 1502 is an entity of the clause 803 in Equation 1. A direction 1503 indicates a transition relationship between the object 1501 and the object 1502.

  As described above, the first embodiment of the present invention has been described until the transition relation of the sentence 801 in the expression 1 is displayed. However, the description and the drawings constituting a part of this disclosure do not limit the present invention. Absent.

DESCRIPTION OF SYMBOLS 100 Design support apparatus 101 Structure data generation part 102 Time / information extraction part 103 Time condition extraction part 104 State transition extraction part 110 TC dictionary 111 sentence DB
112 State transition ID correspondence data 120 Text DB
1200 TC dictionary 1300 sentence DB configuration 1400 state transition data example 1500 execution result image

Claims (5)

A method of extracting the order of sentences using a computer,
A processor included in the computer divides a sentence constituting the sentence into sentences based on a time condition word indicating a transition direction of a processing procedure and a punctuation mark with respect to a sentence to be analyzed. A storage unit having the computer with structure data including a first identifier indicating, a second identifier indicating a sentence including the clause, and, if the clause includes a time condition word, the time condition word A structure data generation step to be stored in
The processor extracts a time condition word from the structure data, and includes time condition data, a transition direction corresponding to the time condition word, and a first identifier corresponding to the time condition word. A time condition extraction step to create
The processor extracts a time condition word from the time condition data, and extracts a first identifier corresponding to a clause having a second identifier identical to the second identifier corresponding to the time condition word from the structure data And, for the second identifier, the first identifier corresponding to the time condition word, the first identifier of the clause having the second identifier, and the time direction corresponding to the time condition word, A sequence extraction step for creating motion sequence data including;
An order extraction method.
The order extraction method according to claim 1,
further,
The processor extracts a character string (state transition word) in which a state transition direction is determined in advance for the state transition between the nodes having the same second identifier, and the first of the clauses including the state transition word A state transition extraction step of creating state transition ID correspondence data including an identifier, a second identifier, and the state transition direction;
An order extraction method.
For the sentence to be analyzed, the sentence constituting the sentence is segmented into sections based on a time condition word indicating the transition direction of the processing procedure and a punctuation mark, and each section includes a first identifier indicating the section and the section; A structure analysis step of creating structure data including a second identifier indicating a sentence including the clause, and, if the clause includes a time condition word, the time condition word;
Time for extracting time condition words from the structure data and creating time condition data including the time condition words, a transition direction corresponding to the time condition words, and a first identifier corresponding to the time condition words A condition extraction step;
Extracting a time condition word from the time condition data; extracting a first identifier corresponding to a clause having a second identifier identical to a second identifier corresponding to the time condition word from the structure data; Operation sequence data including, for two identifiers, a first identifier corresponding to the time condition word, the first identifier of a section having the second identifier, and a time direction corresponding to the time condition word An order extraction step to create a
An order extraction program.
The order extraction program according to claim 3,
further,
For state transitions between clauses having the same second identifier, a character string (state transition word) having a predetermined state transition direction is extracted, and the first identifier and the second for the clause including the state transition word are extracted. A state transition extraction step of creating state transition ID correspondence data including the identifier of the state and the state transition direction;
An order extraction program.
For the sentence to be analyzed, the sentence constituting the sentence is segmented into sections based on a time condition word indicating the transition direction of the processing procedure and a punctuation mark, and each section includes a first identifier indicating the section and the section; A second identifier indicating a sentence including the clause, and, if the clause further includes a time condition word, a structure data generation unit that generates structure data including the time condition word;
Time for extracting time condition words from the structure data and creating time condition data including the time condition words, a transition direction corresponding to the time condition words, and a first identifier corresponding to the time condition words A condition extraction unit;
Extracting a time condition word from the time condition data; extracting a first identifier corresponding to a clause having a second identifier identical to a second identifier corresponding to the time condition word from the structure data; Operation sequence data including, for two identifiers, a first identifier corresponding to the time condition word, the first identifier of a section having the second identifier, and a time direction corresponding to the time condition word An order extractor for creating
A sequence extraction device.