JP2013101509A - Generation device, generation method, and generation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically find out a candidate for a pair of a substance name and its chemical formula in an electronic document.SOLUTION: A generation device extracts a character string "ethane (H3CH3)" estimated to have its substance name and rational formula described in synonymous expression in document information 100-1. The generation device extracts a word "ethane" right before the parentheses in the character string "ethane (H3CH3)". The generation device determines whether or not the word "ethane" is included in a substance name DB in which substance names, and characters and character strings related to the substance names are registered. When the word "ethane" is included, the generation device specifies the word "ethane" as a substance name, and specifies the character string "CH3CH3" having the English letters in the parentheses as a rational formula. The generation device registers the specified substance name "ethane" and rational formula "CH3CH3" in a correspondence relation DB 200.

Description

  The present invention relates to a generation device, a generation method, and a generation program that generate information.

  The prior art which draws a chemical structural formula from a substance name using the naming rule of a chemical substance name is disclosed. The prior art assumes that parts for drawing are registered in a dictionary. The prior art is drawn assuming the ending “ene” indicating hydrogen deficiency and “di” indicating that there are two locations.

Japanese Patent Laid-Open No. 01-142869

  However, the number of existing substances is enormous, and there is a problem in the prior art that it is difficult to prepare a database in which substance names and chemical formulas of existing substances are registered.

  An object of the present invention is to provide a generation apparatus, a generation method, and a generation program capable of automatically finding a candidate for a combination of a substance name and its chemical formula from an electronic document in order to solve the above-described problems caused by the conventional technology. And

  In order to solve the above-described problems and achieve the object, according to one aspect of the present invention, a character string that satisfies a condition representing a set of a substance name and a chemical formula of the substance name is extracted from document information, A generation device, a generation method, and a generation program for registering a set of the substance name and the chemical formula included in the extracted character string in a first database are proposed.

  According to one aspect of the present invention, it is possible to automatically find a candidate for a combination of a substance name and its chemical formula from an electronic document.

FIG. 1 is an explanatory diagram illustrating an example of registration of the correspondence DB by the generation device. FIG. 2 is an explanatory diagram illustrating an example of the contents stored in the correspondence DB 200. FIG. 3 is an explanatory diagram illustrating an example of generating a sexual expression by the generation device. FIG. 4 is an explanatory diagram of an example of the contents stored in the rule DB. FIG. 5 is an explanatory diagram of an example of generating a sexual expression by the generator (part 1). FIG. 6 is an explanatory diagram of an example of generating a sexual expression by the generating device (part 2). FIG. 7 is an explanatory diagram of an example (part 3) of generating a sexual expression by the generation device. FIG. 8 is an explanatory diagram showing an example of the contents stored in the stem dictionary. FIG. 9 is an explanatory diagram showing an example of the contents stored in the partial characterization formula DB. FIG. 10 is a block diagram of a hardware configuration example of the generation apparatus according to the embodiment. FIG. 11 is a block diagram illustrating a functional configuration of the generation device. FIG. 12 is an explanatory diagram showing an example regarding the division of the extracted substance name and the update of the certainty factor. FIG. 13 is an explanatory diagram showing another example regarding the division of the extracted substance name and the update of the certainty factor. FIG. 14 is an explanatory diagram showing an example of increasing the registration number of the partial characteristic formula DB 900. FIG. 15 is a flowchart illustrating a creation processing procedure of the correspondence DB 200 and the partial characteristic formula DB 900 by the generation device 1100. FIG. 16 is a flowchart showing a detailed description of the characteristic type componentization processing (step S1506) shown in FIG. FIG. 17 is a flowchart showing a detailed description of the characteristic expression componentization rule application processing (step S1602) shown in FIG. FIG. 18 is a flowchart showing a certainty factor update processing procedure by the generation device 1100. FIG. 19 is a flowchart showing a detailed description of the extracted substance name dividing process (step S1803) shown in FIG. FIG. 20 is a flowchart showing a detailed description of the registration process (step S1908) shown in FIG. FIG. 21 is a flowchart illustrating an example of a registration processing procedure of the partial referential expression DB 900 by the generation device 1100.

  Exemplary embodiments of a generation device, a generation method, and a generation program according to the present invention will be described below in detail with reference to the accompanying drawings. Further, in this specification, the “chemical formula” is a formula in which a substance is expressed by atomic symbols and numbers. For example, in the case of ethanol, it is C2H6O. The “indicative formula” is a formula in which chemical formulas are further grouped into functional groups. For example, in the case of ethanol, it is CH3CH2OH. Hereinafter, in the present embodiment, a description will be given by taking an illustrative formula as an example, but the present invention can also be applied to a chemical formula.

  The number of substances is enormous, and it is difficult to prepare a dictionary that covers substance names and chemical formulas. In addition, although the chemical substance database in which the substance names and chemical formulas of existing substances are registered is sold, it is expensive, so a new chemical formula can be obtained from the substance names and chemical formulas using the chemical substance database sold. To create it increases the cost.

  Therefore, the generation apparatus according to the present embodiment extracts a part in which a substance name and a chemical formula are presumed to be described in a synonymous expression from an electronic document, and accumulates the combination of the substance name and the chemical formula in a database. Thereby, the generation apparatus can automatically find a candidate for a combination of a substance name and its chemical formula from an existing electronic document.

  In addition, the generation apparatus uses the combination of the substance name and the characteristic expression in the first database created from the document information as a generation source, and applies a chemical reaction rule to newly add the substance name and its characteristic expression. A set may be generated and stored in the second database. In this way, since the generating device generates the referential expression after applying the rules according to the chemical reaction, the substance name and the referential expression are determined without depending on the first database of chemical substances. It is possible to generate the second database with high accuracy and automatically.

  In addition, the generation apparatus determines whether or not the combination of the substance name registered in the first database and the characteristic expression thereof can be reproduced by connecting the substance name and the characteristic expression combination in the second database. You may decide to judge. As described above, the generation device extracts a portion estimated to be described in the synonymous expression from the electronic document, and accumulates the combination of the substance name and the sexual expression in the first database. The part presumed to be described in the synonymous expression is not always accurate.

  For example, parenthesis is mentioned as a synonymous expression, but it is not clear in parenthesis that a description other than a synonym is included, and where from where to what in parenthesis corresponds as a synonymous expression. Therefore, the substance name and chemical formula extracted from the electronic document are not always accurate, and the chemical reaction rules are applied to the substance name and chemical formula in the first database accumulated by extracting from the electronic document. The substance names and chemical formulas obtained in are not always accurate. In addition, for example, when there is sufficient electronic information that serves as an information source for a combination of a substance name and its characteristic formula, it is possible to store a highly accurate set in the first database based on the number of appearances of the set. However, there are not necessarily many opportunities for the synonymous relationship between substance names and their sexual formulas to be described in electronic information.

  Therefore, as described above, the generation device reproduces the combination of the substance name registered in the first database and its characteristic expression by connecting the combination of the substance name and its characteristic expression in the second database. By determining whether or not it can be performed, it is possible to determine the validity of the combination of the substance name registered in the second database and the expression formula thereof. Specifically, for example, the generation device may calculate the accuracy of the combination of the substance name and the characteristic formula in the second database based on the determination result of whether or not the reproduction can be performed. Thereby, this accuracy can be used as an index such as which chemical formula in the second database is useful or which chemical formula may exist.

  Further, the generation apparatus may register the element in the second database if the element of the substance name and the characteristic formula in the first database is not in the second database. As a result, the number of registered types in the second database can be increased, and the combination of the substance name in the first database and its characteristic expression is changed to the combination of the substance name in the second database and its characteristic expression. It is possible to increase the probability of reproducing by connecting.

  In addition, the generation apparatus alternately repeats the process of determining whether or not the combination of the substance name registered in the first database and the characteristic formula can be reproduced and the process of increasing the number of registered types in the second database. By this, it is possible to increase the chances of judging the validity of the combination of the substance name registered in the second database and its expression formula.

<Example of Registration of Correspondence DB (Data Base)>
First, an example will be described in which the generation apparatus extracts a part in which a substance name and a chemical formula are presumed to be described in synonymous expressions from an electronic document, and accumulates the combination of the substance name and the chemical formula in a database.

  FIG. 1 is an explanatory diagram illustrating an example of registration of the correspondence DB by the generation device. The document information DB 100 is a database that stores document information including texts such as patent documents, papers, magazines, books, and the like relating to chemistry and pharmacy. The document information may be the entire electronic document or information obtained by extracting a part from the electronic document. Taking a patent document as an example, the document information may be information obtained by extracting a description portion related to the invention in the patent document. In FIG. 1, the document information DB 100 has a plurality of document information 100-1 to 100-n. The document information DB 100 is realized by a storage device such as a ROM (Read-Only Memory), a RAM (Random Access Memory), a flash memory, a disk, or the like included in the generation device.

  The correspondence DB 200 is the first database described above. The correspondence DB 200 includes character strings that satisfy a condition that represents a set of a substance name and a characteristic expression among document information 100-i (i = 1 to n) selected from the document information DB 100 by the generation device. The specified substance name and indication formula are registered.

  The character string that satisfies the condition that represents the combination of the substance name and the expression of the substance name is, for example, a character string that includes a synonymous expression. For example, since the parenthesis and the word immediately before the parenthesis may be synonymous, the character string including the synonymous expression is, for example, a sentence having parentheses. Examples of parentheses include “”, (), {}, and [].

  For example, the generation device extracts a sentence including parentheses from the document information. Then, for example, the generation device identifies a word immediately before the parentheses from a sentence including the parentheses as a substance name and a character string in the parentheses as a referential expression.

  Further, for example, a database in which a substance name and characters representing the substance name are recorded may be prepared in advance. Here, this database is referred to as a substance name DB. Examples of the character representing the substance name include “acid”, “group”, “element”, “salt”, “charcoal”, and the like. For example, the generation apparatus determines whether the word immediately before the parenthesis includes a character string in the substance name DB. When the word includes a character string in the substance name DB, the generation apparatus specifies the word as the substance name and specifies the character string in parentheses as a referential expression.

  The generation device may determine whether or not an alphabet is included in the character string in the parentheses. If the character string in parentheses does not contain alphabets, the generator does not identify the character string in parentheses as a descriptive expression, and if the character strings in parentheses contain alphabets, the generator The character string in parentheses is specified as an indicia expression.

  Then, the generation apparatus registers the specified substance name and characteristic formula pair in the correspondence DB 200. The correspondence DB 200 will be described in detail with reference to FIG.

  Taking the document information 100-1 as an example, the generation apparatus extracts “ethane (H3CH3)”, “ethanesulfonic acid“ CH3CH2SO3H ”, and“ ethanesulfonic acid [CHES] ”as a sentence including parentheses.

  For example, the generation apparatus determines whether or not the word “ethane” immediately before the parentheses in “ethane (H3CH3)” is included in the substance name DB. When the word “ethane” is included in the substance name DB, the generating apparatus identifies the word “ethane” as the substance name. Then, the generation device determines whether or not the character string “CH3CH3” in parentheses is a character string including alphabets. When the character string “CH3CH3” in the parentheses is a character string including an alphabet, the generation device specifies the character string “CH3CH3” in the parentheses as a sexual expression. The generation device may determine whether or not the character string in parentheses is a character string including an alphabet. Then, the generation apparatus registers the specified combination of the substance name and the characteristic formula in the correspondence DB 200.

  Similarly, for example, the generation device specifies the character string “CH3CH2SO3H” in parentheses as a sexual expression with the word “ethanesulfonic acid” immediately before the parentheses in “ethanesulfonic acid“ CH3CH2SO3H ”as a substance name. . Similarly, the generation apparatus identifies “ethanesulfonic acid” in “ethanesulfonic acid [CHES]” as a substance name and “CHES” in parentheses as a sexual expression. Then, the generation apparatus registers the specified combination of the substance name and the characteristic formula in the correspondence DB 200.

  Further, the generation device determines whether the character string in the parenthesis is included in the substance name DB, thereby specifying the character string in the parenthesis as the substance name, and the word immediately before the parenthesis May be specified.

  Moreover, as description of synonymous expression, "CH3SH is methanethiol" is mentioned in the document information 100-1, for example. For example, the generation device identifies a word “methanethiol” including a character string registered in the substance name DB as a substance name among “CH3SH is methanethiol”. For example, the generation device specifies “CH3SH” including an alphabet as a referential expression. Then, the generation apparatus registers the specified combination of the substance name and the characteristic formula in the correspondence DB 200.

  According to FIG. 1, the generation apparatus can automatically find a candidate for a combination of a substance name and its chemical formula from an existing electronic document.

  As described above, according to the generation apparatus, a candidate for a combination of a substance name and its chemical formula can be automatically found from an existing electronic document and can be accumulated in the correspondence DB 200. Next, the stored contents of the correspondence DB 200 in which a set of the substance name extracted by the generation device and its characteristic formula is registered will be described with reference to FIG.

<Correspondence DB 200>
FIG. 2 is an explanatory diagram illustrating an example of the contents stored in the correspondence DB 200. The correspondence relationship DB 200 is a database that stores the substance names extracted from the document information 100-1 to 100-n in the document information DB 100 by the generating device and their characteristic formulas in association with each other. The correspondence DB 200 includes a substance name item and a referential expression item. In the correspondence DB 200, each item is set for each substance name to form a record. The substance name is registered in the substance name item. In addition, the indicia formula is registered in the indicia formula item.

  For example, the initial state of the correspondence DB 200 is empty, and is added to the record every time a set of a substance name and its characteristic formula is generated. FIG. 2 shows a state where substance names and sexual formulas have already been registered for substances such as methane and methoxy. The combination of the substance name and the characteristic formula defined in the correspondence DB 200 is selected from the generation device when generating the partial characteristic expression.

  The correspondence DB 200 is realized, for example, by a storage device such as a ROM, a RAM, a flash memory, or a disk included in the generation device.

<Generation formula generation example>
FIG. 3 is an explanatory diagram illustrating an example of generating a sexual expression by the generation device. In FIG. 3, “≈” indicates the correspondence between the substance name and the sexual expression. In FIG. 3, as an example, a combination of the substance name “propane” and the characteristic expression “CH3CH2CH3” is used as a generation source among the substance name and characteristic expression pairs included in the correspondence DB 200.

In (1), the rule that “hydrogen is removed as a chemical reaction when a hydrocarbon CH is bonded” is applied to the substance name “propane” and the characteristic formula “CH 3 CH 2 CH 3”. In the propane characteristic formula “CH 3 CH 2 CH 3”, there are 3 hydrogen Hs (total 8), and the rule can be applied to any of the sites, but here, for convenience of explanation, from the last “CH 3” to the hydrogen H Suppose that one is removed. As a result, the substance name “propane” and the characteristic formula “CH3CH2CH2” are obtained. In the present specification, the applied expression after application of the rule is referred to as “partial expression”. In addition, the substance name after the application of the rule is referred to as “partial name”.

In (2), “hydrogen H is replaced by“ OH ”and“ nor ”is added to the end of the partial name for the partial name“ propane ”and the partial characteristic formula“ CH3CH2CH2 ”obtained in (1). The rule is applied. In partial partial formula “CH3CH2CH2”, there are three hydrogen Hs (a total of seven), but here, it is assumed that hydrogen H in the center “CH2” is replaced with OH. Thereby, the partial characteristic formula “CH3CHOHCH2” is obtained. In addition, “nor” is added to the end of the partial name “propane”, but in the case of adding the end, the end “nor” is added to the stem “propa”, and the partial name “propanol” is obtained.

In (3), “hydrogen H is replaced by fluorine“ F ”and“ fluoro ”is added to the beginning of the partial name for the partial name“ propane ”and the partial characteristic formula“ CH3CH2CH2 ”obtained in (1). The rule that “does” is applied. In the partial characteristic formula “CH3CH2CH2”, there are three hydrogens H (a total of seven). Here, it is assumed that the hydrogen “H” in the leading “CH3” is replaced with fluorine F. Thereby, the partial characteristic formula “CH2FCH2HCH2” is obtained. Further, “fluoro” is added to the beginning of the partial name “propane” to obtain the partial name “fluoropropane”.

In (4), “hydrogen H is replaced by chlorine Cl and“ chloro ”is added to the beginning of the partial name for the partial name“ propanol ”and the partial characteristic formula“ CH 3 CHOHCH 2 ”obtained in (2). Is applied. In the partial characteristic formula “CH3CHOHCH2”, there are four hydrogen Hs (total of seven), but here, it is assumed that the hydrogen H of the leading “CH3” is replaced by chlorine Cl. Thereby, the partial characteristic formula “CH 2 ClCHOHCH 2” is obtained. In addition, “chloro” is added to the beginning of the partial name “propanol” to obtain the partial name “chloropropanol”.

In (5), “hydrogen H is replaced with fluorine F and“ fluoro ”is added to the beginning of the partial name for the partial name“ propanol ”and the partial characteristic formula“ CH3CHOHCH2 ”obtained in (2). Is applied. The rule is applied in (3) above, but the same rule may be applied multiple times. In the partial characteristic formula “CH 3 CHOHCH 2”, there are four hydrogen H positions (total of seven). Here, it is assumed that the hydrogen H in the leading “CH 3” is replaced with fluorine F. As a result, the partial characteristic formula “CH2FCHOHCH2” is obtained. Further, “fluoro” is added to the beginning of the partial name “propanol” to obtain the partial name “fluoropropanol”.

In (6), “hydrogen H is replaced by fluorine F and“ fluoro ”is added to the beginning of the partial name for the partial name“ fluoropropanol ”obtained in (5) and the partial characteristic formula“ CH 2 FCHOHCH 2 ”. The rule is applied. In this case, when the hydrogen “H” in the leading “CH2” is replaced by fluorine F, the partial name and partial characteristic formula after generation are the partial partial expression “CHClFCHOHCH2” of the generated partial name “chlorofluoropropanol”. Will match. In such a case, the generation process is stopped. Therefore, the generation device can suppress the duplicate generation of the part name and the partial expression, and can reduce the load of the generation process.

  As described above, according to the generating device, the expression is generated after applying the rule according to the chemical reaction. Therefore, the substance name and the indicating characteristic are not dependent on the first database of the chemical substance. The expression can be generated with high accuracy and the second database can be automatically created.

  Next, the rules applied in FIG. 3 will be described. In the present embodiment, a rule is information that defines a condition relating to a chemical reaction to be applied and contents of conversion when the condition is satisfied. A rule group that is a set of rules is stored in the rule DB. The rule DB is realized, for example, by a storage device such as a ROM, a RAM, a flash memory, or a disk included in the generation device.

<Rule DB>
FIG. 4 is an explanatory diagram of an example of the contents stored in the rule DB. The rule DB 400 includes a rule ID item, an application condition item, and a conversion content item. The rule DB 400 configures a record in which each item is set for each rule. The rules defined in the rule DB 400 are selected from the generating device when generating the partial expression.

  In the rule ID item, the rule ID is stored for each record. The rule ID is information that uniquely identifies a rule. In the application condition item, an application condition is stored for each record. The application condition is a condition relating to a chemical reaction that should be satisfied by the characteristic expression or the partial characteristic expression in order to apply the rule. For example, since the ethanol equation is “CH3CH2OH”, the application condition of rule ID = R1 is satisfied.

  The conversion content item includes a referential expression item and a substance name item. The chemical formula applied to the differential formula or the partial differential formula is stored for each record in the differential formula item. In the substance name item, name change contents applied to the substance name and the partial name are stored for each record. There are three types of name change contents: “no change”, “addition of prefix”, and “addition of ending”.

  FIG. 5 is an explanatory diagram of an example of generating a sexual expression by the generator (part 1). FIG. 5 shows an example in which the rule ID = R1 of the rule DB 400 shown in FIG. 1 is applied. In FIG. 5, the generation source is assumed to be the substance name “propane” and the characteristic formula “CH3CH2CH3”.

  When the rule R1 is applied, three types of partial characteristic expressions “CH2CH2CH3”, “CH3CHCH3”, and “CH3CH2CH2” are generated from the characteristic expression “CH3CH2CH3”. Further, the substance name “propane” is not changed, and “propane” becomes a partial name as it is.

  FIG. 6 is an explanatory diagram of an example of generating a sexual expression by the generating device (part 2). FIG. 6 shows an example in which the rule ID = R2 of the rule DB 400 shown in FIG. 4 is applied. In FIG. 6, the generation source is a partial name “propane” and a partial characteristic formula “CH2CH2CH3”.

  When the rule R2 is applied, three types of partial characteristic expressions “CHOHCH2CH3”, “CH2CHOHCH3”, and “CH2CH2CH2OH” are generated from the partial characteristic expression “CH2CH2CH3”. In addition, the partial name “propane” becomes the partial name “propanol” for any partial characteristic formula. If the ending “nor” is added to the partial name “propane” as it is, it will be “propanol”, but in the case of adding the ending, the ending “nor” will be added to the stem “propa” of “propane”. Becomes “propanol”.

  FIG. 7 is an explanatory diagram of an example (part 3) of generating a sexual expression by the generation device. FIG. 7 shows an example in which the rule ID = R3 of the rule DB 400 shown in FIG. 4 is applied. In FIG. 7, the generation source is a partial name “propanol” and a partial characteristic formula “CHOHCH2CH3”.

  When the rule R3 is applied, the four partial partial expressions of “CClOHCH2CH3”, “CHOClCH2CH3”, “CHOHCHClCH3”, and “CHOHCH2CH2Cl” are generated from the partial incremental expression “CHOHCH2CH3”. In addition, the partial name “propane” becomes the partial name “chloropropanol” for any partial referential formula. Next, the stored contents of the stem dictionary that is referred to by the generating device when a ending is added in the conversion contents of the rule will be described with reference to FIG.

<Stem stem dictionary>
FIG. 8 is an explanatory diagram showing an example of the contents stored in the stem dictionary. The stem dictionary 800 is a database for specifying the stems of substance names and partial names. In the word stem dictionary 800, word stems with known substance names are registered. When a ending is added in the conversion contents of the rule, the generating apparatus refers to the stem dictionary 800. Then, the generation device searches the stem dictionary 800 for the longest match, extracts the longest match stem, and adds the ending specified in the applied rule to the end of the extracted stem.

  For example, in the example of FIG. 6, for the partial name “propane”, “propa” has the longest match, so the stem “propa” is read from the stem dictionary 800. Then, the ending “nor” is added to the read stem “prop” and becomes “propanol”. The stem dictionary 800 is realized by a storage device such as a ROM, a RAM, a flash memory, or a disk included in the generation device, for example. Next, with reference to FIG. 9, a description will be given of the stored contents of the partial characteristic formula DB in which a set of the partial name generated by the generation device and the partial characteristic formula is registered.

<Partial expression formula DB>
FIG. 9 is an explanatory diagram showing an example of the contents stored in the partial characterization formula DB. The partial characteristic formula DB 900 is a database for registering a part name generated by the generation device and a set of the partial characteristic formulas. The partial characteristic expression DB 900 includes a partial name item, a partial characteristic expression item, and a certainty factor item. In the partial characteristic formula DB 900, each item is set for each substance specified by the partial name, and constitutes a record. The partial name is registered in the partial name item. In addition, a partial characteristic expression is registered in the partial characteristic expression item. In the certainty factor item, a certainty factor indicating the certainty of the set of the partial name registered in the partial name item and the partial indicia equation registered in the partial intensity formula item is registered. That is, the certainty factor represents the certainty that the partial characteristic expression is the partial name characteristic expression. In the present embodiment, the certainty factor is represented by a score, and the certainty factor is high in the descending order of the score. However, the present invention is not limited to this, and the certainty factor may be high in the descending order of the score. Further, for example, the certainty factor may be represented by a rank. For example, when the generation device updates the certainty level to a value indicating more certainty, when the certainty level is rank, the rank is increased, and when the certainty level is high and the high level is high, increase.

  For example, the initial state of the partial characteristic formula DB 900 is empty, and is added to the record every time a set of a partial name and the partial characteristic formula is generated. When a set of a partial name and its partial characteristic formula is added to the record, 0 is set in the certainty factor item.

  FIG. 9 shows a state where partial names and partial sexual formulas have already been registered for substances such as methane and methoxy.

  In addition, when the partial name and the combination of the partial characteristic formula are registered in the partial characteristic formula DB 900, the generation apparatus searches the partial characteristic expression DB 900 for the combination of the partial name and the partial characteristic formula. be able to. For example, patent documents and academic papers may describe new substance names and their formulas, but the generator stores such new substance names and their formulas in the chemical database. Even if it is not done, it can be searched from the database generated by the generation device.

<Example of hardware configuration of generation device>
FIG. 10 is a block diagram of a hardware configuration example of the generation apparatus according to the embodiment. In FIG. 10, the generation apparatus includes a CPU (Central Processing Unit) 1001, a ROM 1002, and a RAM (Random Access Memory) 1003. The generation apparatus includes a magnetic disk drive 1004, a magnetic disk 1005, an optical disk drive 1006, an optical disk 1007, a display 1008, an I / F (Interface) 1009, a keyboard 1010, and a mouse 1011. . Each unit is connected by a bus 1000.

  Here, the CPU 1001 governs overall control of the generation apparatus. The ROM 1002 stores a program such as a boot program. The RAM 1003 is used as a work area for the CPU 1001. The magnetic disk drive 1004 controls reading / writing of data with respect to the magnetic disk 1005 according to the control of the CPU 1001. The magnetic disk 1005 stores data written under the control of the magnetic disk drive 1004.

  The optical disc drive 1006 controls reading / writing of data with respect to the optical disc 1007 according to the control of the CPU 1001. The optical disc 1007 stores data written under the control of the optical disc drive 1006, and causes the computer to read data stored on the optical disc 1007.

  The display 1008 displays data such as a document, an image, and function information as well as a cursor, an icon, or a tool box. As the display 1008, for example, a CRT, a TFT liquid crystal display, a plasma display, or the like can be adopted.

  The I / F 1009 is connected to a network 1014 such as a LAN (Local Area Network), a WAN (Wide Area Network), and the Internet through a communication line, and is connected to another device via the network 1014. The I / F 1009 controls an internal interface with the network 1014 and controls input / output of data from an external device. For example, a modem or a LAN adapter may be employed as the I / F 1009.

  The keyboard 1010 includes keys for inputting characters, numbers, various instructions, and the like, and inputs data. Moreover, a touch panel type input pad or a numeric keypad may be used. The mouse 1011 performs cursor movement, range selection, window movement, size change, and the like. A trackball or a joystick may be used as long as they have the same function as a pointing device.

<Functional configuration example of generation device>
FIG. 11 is a block diagram illustrating a functional configuration of the generation device. In FIG. 11, the generation apparatus 1100 includes a document information DB 100, a substance name DB 1120, a correspondence relationship DB 200, a rule DB 400, a stemming dictionary 800, and a partial characteristic formula DB 900. In the substance name DB 1120, characters and character strings related to substance names and substance names are registered. The substance name DB 1120 is realized by a storage device such as the ROM 1002, the RAM 1003, the magnetic disk 1005, and the optical disk 1007, for example.

  The generation apparatus 1100 includes a character string extraction unit 1101, a first registration unit 1102, a rule extraction unit 1103, a determination unit 1104, a conversion unit 1105, a second registration unit 1106, and a determination unit 1107. ,have. The generation apparatus 1100 includes an acquisition unit 1108, a division unit 1109, a search unit 1110, a generation unit 1111, a determination unit 1112, an update unit 1113, and a deletion unit 1114. Specifically, the character string extraction unit 1101 to the deletion unit 1114 causes the CPU 1001 to execute a generation program stored in a storage device such as the ROM 1002, the RAM 1003, the magnetic disk 1005, and the optical disk 1007 shown in FIG. Alternatively, the function is realized by the I / F 1009. Further, the processing results of the respective functional units are stored in a storage device such as the RAM 1003, the magnetic disk 1005, and the optical disk 1007, for example.

  The character string extraction unit 1101 extracts a character string satisfying a condition representing a set of a substance name and a substance name expression formula from the document information. As described above, the character string that satisfies the condition representing the combination of the substance name and the explicit expression of the substance name is, for example, a character string that includes a synonymous expression. For example, since the parenthesis and the word immediately before the parenthesis may be synonymous, the character string including the synonymous expression is, for example, a sentence having parentheses. Specifically, for example, the character string extraction unit 1101 extracts a sentence having parentheses from the document information in the document information DB 100.

  For example, as shown in FIG. 1, “ethane (H3CH3)” having parentheses is extracted from the document information 100-1. The character string extraction unit 1101 extracts the word “ethane” immediately before the parenthesis from “ethane (H3CH3)”. Then, the character string extraction unit 1101 searches the substance name DB 1120 to determine whether or not a partial character string of the word “ethane” is included in the substance name DB 1120. When the partial character string of the word “ethane” is included, the character string extraction unit 1101 identifies the word “ethane” as a substance name and identifies the character string “CH3CH3” in parentheses as a sexual expression.

  Then, the first registration unit 1102 registers a set of the substance name included in the extracted character string and its characteristic formula in the database. Specifically, for example, the first registration unit 1102 registers a set of a substance name and its characteristic formula in the correspondence DB 200. For example, a set of the specified substance name “ethane” and the characteristic formula “CH3CH3” is registered in the correspondence DB 200.

  The rule extraction unit 1103 extracts any rule from the rule group that defines the condition to be applied and the conversion content when the condition is satisfied. Specifically, for example, the rule extraction unit 1103 extracts any rule from the rule DB 400. The rule extraction unit 1103 may extract random rules or may be in the order of rule IDs. The extracted rules may be duplicated.

  The determination unit 1104 determines whether or not the characteristic expression registered by the first registration unit 1102 satisfies a condition to be applied in any of the rules extracted by the rule extraction unit 1103. Specifically, for example, the determination unit 1104 determines whether or not the characteristic expression registered by the first registration unit 1102 satisfies the application condition of the rule extracted from the rule DB 400. For example, in the examples of FIGS. 5 to 7, the application conditions of the extracted rules R1 to R3 are satisfied. On the other hand, when the rule of the application condition (see FIG. 3) of “bonding hydrocarbon CH” is extracted for ethanol, the application condition is not satisfied.

  When it is determined that the condition is satisfied by the determination unit 1104, the conversion unit 1105 converts the referential expression and the substance name according to the conversion content. Specifically, for example, when the conversion unit 1105 determines that the characteristic expression satisfies the application condition of the extracted rule, the conversion unit 1105 converts the substance name and the characteristic expression according to the conversion content of the extracted rule. And output the part name and its partial expression. Also, the converted part name and the partial characteristic formula are fed back to the rule extraction unit 1103. As a result, the extracted processing by the rule extraction unit 1103, the determination processing by the determination unit 1104, and the conversion processing by the conversion unit 1105 are recursively executed for the converted partial name and the partial characteristic expression.

  In addition, about the conversion example by the conversion part 1105, as shown in FIGS. 5-7, the conversion according to a chemical reaction is performed for a characteristic formula or a partial characteristic formula. For the substance name or partial name, no change, addition of a prefix, or addition of a ending is executed. When addition of the ending is defined in the extracted rule, the conversion unit 1105 extracts the stem of the substance name or the partial name by the longest match search with respect to the stem dictionary 800. Then, the conversion unit 1105 generates a new partial name by adding a rule-defined ending to the extracted word stem.

  In addition, as illustrated in FIGS. 5 to 7, the conversion unit 1105 obtains a plurality of conversion candidates. However, any conversion candidate may be converted, or any one may be converted. . That is, at least one conversion may be performed. When there are a plurality of conversion candidates, the user may set an upper limit of the number to be converted in advance. For example, when the user wants to realize the construction of the partial characteristic formula DB 900 at an early stage, the user only has to set the upper limit to a low value (for example, upper limit = 1). The user may set the upper limit higher.

  The second registration unit 1106 registers the converted chemical formula and substance name in the database. Specifically, for example, the second registration unit 1106 registers the partial name that is the conversion result and its characteristic expression in the partial characteristic expression DB 900. Thereby, as shown in FIG. 9, the partial characteristic formula DB 900 is constructed. Note that if the partial name as a conversion result and its characteristic expression have already been registered in the partial characteristic expression DB 900, the second registration unit 1106 does not execute the registration process. Thereby, the increase in the number of records is suppressed and the memory saving of the partial characteristic formula DB 900 can be achieved.

  The determination unit 1107 determines whether the converted chemical formula and substance name are those newly registered in the second registration unit 1106. Specifically, for example, the determination unit 1107 determines whether or not the registration of the partial name that is the conversion result and the specific expression in the partial characteristic formula DB 900 is a new registration. In the case of new registration, the rule extraction unit 1103 extracts one of the rules from the rule group. That is, the conversion result is fed back from the conversion unit 1105 to the rule extraction unit 1103, but the rule is extracted from the rule DB 400 only when the determination unit 1107 determines new registration. As a result, the extraction process, the determination process, and the conversion process for the registered partial name and its partial characteristic formula are suppressed, and the efficiency of the construction work of the partial characteristic formula DB 900 can be improved.

  Next, the generation apparatus 1100 determines whether or not the combination of the substance name in the correspondence DB 200 and its characteristic formula can be reproduced by connecting the substance name in the partial characteristic formula DB 900 and the combination of the specific expression. A process for determining the accuracy of the combination of the substance name registered in the partial characteristic formula DB 900 and the characteristic expression by making the determination will be described.

  The acquisition unit 1108 acquires the substance name extracted from the document information and the chemical formula of the substance name. Specifically, for example, the acquisition unit 1108 selects a record having the correspondence DB 200, and reads a set of a substance name and an explicit expression of the substance name in the selected record. Here, the substance name in the read record is the extracted substance name, and the characteristic expression in the selected record is the extraction characteristic.

  Then, the dividing unit 1109 extracts the first substance name in the database that matches the extracted substance name from the extracted substance name, and extracts the first substance name from the extracted substance name from the remaining character string. By extracting the second substance name in the database that matches the remaining character string, the extracted substance name is divided into the first substance name and the second substance name.

  FIG. 12 is an explanatory diagram showing an example regarding the division of the extracted substance name and the update of the certainty factor. FIG. 12 illustrates an example in which the acquisition unit 1108 reads the substance name “ethanesulfonic acid” and the qualitative formula “CH3CH2SO3H” from the correspondence DB 200. Specifically, for example, the dividing unit 1109 searches the partial characteristic formula DB 900 and extracts the partial name whose partial name in the partial characteristic formula DB 900 is the longest match from the beginning of the extracted substance name “ethanesulfonic acid”. To do. In this case, it is assumed that the partial name “ethane” is extracted.

  Similarly, the dividing unit 1109 searches the partial characteristic formula DB 900 for the remaining character string “sulfonic acid” obtained by removing the extracted partial name “ethane” from the extracted substance name “ethanesulfonic acid”. As a result, when the partial name “sulfonic acid” is searched for the longest match, the dividing unit 1109 extracts the partial name “sulfonic acid” from the remaining character string “sulfonic acid”. As a result, since there is no remaining character string, the division by the dividing unit 1109 is successful. Thereby, the dividing unit 1109 can divide the extracted substance name “ethanesulfonic acid” into the partial name “ethane” and the partial name “sulfonic acid”.

  The searching unit 1110 searches the database for the first chemical formula of the first substance name divided by the dividing unit 1109 and searches the database for the second chemical formula of the second substance name divided by the dividing unit 1109. To do. In the example of FIG. 12, the dividing unit 1109 divides the extracted substance name “ethanesulfonic acid” into a partial name “ethane” and a partial name “sulfonic acid”.

  Specifically, the search unit 1110 searches the partial characteristic formula DB 900 for the characteristic expression “CH3CH2” and the characteristic expression “CHF2OCHFCF3” of the partial name “ethane” obtained by the division. In addition, the search unit 1110 searches the partial formula DB 900 for the formula “SO3H” of “sulfonic acid” obtained by the division. Further, when three or more partial names are obtained by the dividing unit 1109, the search unit 1110 performs a search for each partial name.

  The generation unit 1111 generates a connected chemical formula obtained by connecting the first chemical formula and the second chemical formula searched by the search unit 1110. Referring to FIG. 12, the generation unit 1111 includes the characteristic formula “CH3CH2” of the partial name “ethane” searched from the partial characteristic formula DB 900 and “sulfonic acid” searched from the partial characteristic formula DB 900. Link to the formula “SO3H”. The order of concatenation is the order of the part names. Accordingly, the characteristic formula “CH3CH2” of the partial name “ethane” is at the head, and the characteristic formula “SO3H” of “sulfonic acid” is at the end. By this connection, the generation unit 1111 generates “CH3CH2SO3H” as the connection characteristic formula.

  In addition, as shown in FIG. 12, there may be a case where a plurality of characteristic expressions are searched in the search unit 1110. For example, in addition to “CH3CH2”, the partial formula “CHF2OCHFCF3” is retrieved from the partial formula DB900 in addition to “CH3CH2”. In this case, “CHF2OCHFCF3SO3H” is generated in addition to the above-described linkage characteristic formula “CH3CH2SO3H”.

  That is, when m1 partial characteristic expressions are retrieved from one divided partial name and m2 partial characteristic expressions are retrieved from the other partial characteristic expression, m1 × m2 connected characteristic expressions are retrieved. Will be generated. When three or more characteristic expressions are retrieved by the retrieval unit 1110, the generation unit 1111 generates a connected characteristic expression by concatenating the retrieved characteristic expressions. That is, assuming that the number of divided partial names is k and the number of partial characteristic expressions retrieved from each partial name is m1, m2,..., Mk, m1 × m2 ×. Will be generated. When the partial characteristic formula is not obtained by the search, the generation device 1100 treats the character string indicating that it is unknown, for example, “XXXX”. In the above example, when the formula of the sulfonic acid cannot be obtained by the search, “CH3CH2XXXX” is obtained.

  Next, the determination unit 1112 determines whether the linked chemical formula generated by the generation unit 1111 matches the extracted chemical formula that is a set of extracted substance names included in the first database. Specifically, for example, the determination unit 1112 determines whether or not the connection characteristic formulas “CH3CH2SO3H” and “CHF2OCHFCF3SO3H” completely match the extraction characteristic formula “CH3CH2SO3H”. The determination unit 1112 determines that the coupled characteristic “CH3CH2SO3H” matches the extracted characteristic, but the linked characteristic “CHF2OCHFCF3SO3H” does not match the extracted characteristic.

  When it is determined by the determination unit 1112 that the update unit 1113 determines that they match, the update unit 1113 further determines the certainty factor representing the certainty of the first chemical formula and the certainty factor representing the certainty of the second chemical formula included in the second database. Update to a value that indicates the likelihood. Specifically, when it is determined by the determination unit 1112 that the update unit 1113 matches, the update unit 1113 has a certainty factor representing the certainty of the first chemical formula and a certainty factor representing the certainty of the second chemical formula than before the update. Make it high.

  More specifically, for example, when the determination unit 1112 determines that the update unit 1113 matches, the update unit 1113 adds points to the certainty factor in the partial characteristic formula DB 900 related to the partial characteristic formulas that form the connected combined formula. Since it is determined that the combined compound expression “CH3CH2SO3H” matches the extracted characteristic expression, the updating unit 1113 adds one point to the certainty factor of the record having the partial name “ethane” and the partial characteristic expression “CH3CH2”. The updating unit 1113 adds one point to the certainty factor of the record having the partial name “sulfonic acid” and the partial characteristic formula “SO3H”. Here, the number of added points is one point, but is not particularly limited. On the other hand, since it is determined that the coupled characteristic “CHF2OCHFCF3SO3H” does not match the extracted characteristic, the updating unit 1113 does not update the certainty factor of the record having the partial name “ethane” and the partial characteristic “CHF2OCHFCF3”. For example, the update unit 1113 may rearrange the records in the partial characteristic formula DB 900 in descending order of certainty.

  As a result, regardless of whether or not the substance name extracted from the document information and its characteristic expression are incorrect, each time the reproduced substance name and its characteristic expression are reproduced, a portion in the partial characteristic expression DB 900 is reproduced. The certainty of the partial name and the partial characteristic expression can be determined by whether or not the name and the partial characteristic expression contributed.

  In the example of dividing the extracted substance name by the dividing unit 1109 described above, an example in which the first partial name “ethane” and the second partial name “sulfonic acid” are divided has been described. If the character string remains, the longest match search of the partial characteristic formula DB 900 is recursively performed until there is no remaining character string. As a result, when the division is successful, it is divided into three or more partial names, and a characteristic expression is acquired for each obtained partial name.

  Next, another example regarding the division of the extracted substance name and the update of the certainty factor by the acquisition unit 1108 will be described with reference to FIG.

  FIG. 13 is an explanatory diagram showing another example regarding the division of the extracted substance name and the update of the certainty factor. FIG. 13 illustrates an example in which the acquisition unit 1108 reads the extracted substance name “ethanesulfonic acid” and the sexual expression “CHES” from the correspondence DB 200. In FIG. 13, as in the example shown in FIG. 12, the dividing unit 1109 divides the extracted substance name “ethanesulfonic acid” into a partial name “ethane” and a partial name “sulfonic acid”.

  In FIG. 13, similarly to the example illustrated in FIG. 12, the generation unit 1111 generates the coupled characteristic expressions “CH3CH2SO3H” and “CHF2OCHFCF3SO3H”. Then, for example, the determination unit 1112 determines whether or not the linked characteristic expressions “CH3CH2SO3H”, “CHF2OCHFCF3SO3H”, and the extracted characteristic expression “CHES” completely match from the top.

  In the example of FIG. 13, the determination unit 1112 determines that none of the connected characteristic expressions matches the extracted characteristic expression. Then, the update unit 1113 includes the certainty factor of the record including the partial name “ethane” and the partial characteristic expression “CH3CH2”, the certainty factor of the record including the partial name “ethane” and the partial characteristic expression “CHF2OCHFCF3”, and the partial The certainty factor of the record including the name “sulfonic acid” and the partial characteristic formula “SO3H” is not updated.

  In addition, in the example of dividing the extracted substance name by the dividing unit 1109 as shown in FIG. 12 and FIG. 13, the success example has been described, but when the longest match search in the partial characteristic formula DB 900 is not performed, the division fails It becomes. As for the division failure, the case of the substance name “methanethiol” and the characteristic formula “CH3SH” in the correspondence DB 200 will be described as an example. The substance name “methanethiol” is the extraction substance name, and the characteristic formula “CH3SH” is the extraction characteristic formula.

  Specifically, for example, the dividing unit 1109 searches the partial characteristic formula DB 900 and extracts a partial name whose partial name in the partial characteristic formula DB 900 is the longest match from the beginning of the extracted substance name “methanethiol”. . In this case, it is assumed that the partial name “methane” is extracted.

  Next, the dividing unit 1109 similarly searches the partial characteristic formula DB 900 for the remaining character string “thiol” obtained by removing the extracted partial name “methane” from the extracted substance name “methanethiol”. The dividing unit 1109 does not search for the remaining character string “thiol” by the longest match search of the partial characteristic formula DB 900. Therefore, the dividing unit 1109 fails to divide.

  Then, the search unit 1110 searches the partial characteristic formula DB 900 for the partial characteristic formula of the partial name searched by the dividing unit 1109. For example, the search unit 1110 searches the partial characteristic formula DB 900 for the partial characteristic formula “CH3” having the partial name “methane”.

  The second registration unit 1106 stores the combination of the characteristic formula obtained by removing the partial characteristic expression of the partial name searched by the search unit 1110 from the extracted characteristic expression and the remaining character string in the partial characteristic expression DB 900. sign up. Specifically, for example, the second registration unit 1106 includes a characteristic formula “SH” obtained by removing the partial characteristic formula “CH3” from the extracted characteristic formula “CH3SH”, the remaining character string “thiol”, Are registered in the partial characteristic formula DB 900. In the example of “methanethiol”, the partial characteristic formula is included in the extraction characteristic formula. For example, when the partial characteristic formula is not included in the extraction characteristic formula, the second registration unit 1106 Do not register the combination of the referential expression and the substance name.

  Further, an example of newly registering a partial name and a partial characteristic formula in the partial characteristic formula DB 900 from the characteristic formula of the substance name in the correspondence DB 200 will be described with reference to FIG.

  FIG. 14 is an explanatory diagram showing an example of increasing the registration number of the partial characteristic formula DB 900. For example, the acquisition unit 1108 selects one set of the substance name and the characteristic formula from the correspondence DB 200. Here, the substance name “methanethiol” and the characteristic formula “CH3SH” will be described as examples. Specifically, for example, the dividing unit 1109 searches the partial characteristic formula DB 900 and extracts a partial name whose partial name in the partial characteristic formula DB 900 is the longest match from the beginning of the extracted substance name “methanethiol”. . In this case, it is assumed that the partial name “methane” is extracted.

  Similarly, the dividing unit 1109 searches the partial characteristic formula DB 900 for the remaining character string “thiol” obtained by removing the extracted partial name “methane” from the extracted substance name “methanethiol”. As a result, the partial name “thiol” is not searched for the longest match.

  Therefore, the search unit 1110 searches the database for the chemical formula of the partial name divided by the dividing unit 1109. Specifically, for example, the search unit 1110 searches the partial characteristic formula DB 900 for the partial characteristic formula “CH3” of the partial name “methane”.

  The deletion unit 1114 deletes the first character from the remaining character string that has not been searched from the partial characteristic formula DB 900 by the dividing unit 1109. For example, the deletion unit 1114 deletes the first character “h” of the remaining character string “thiol”.

  The dividing unit 1109 searches the partial characteristic formula DB 900 and extracts a partial name in which the partial names in the partial characteristic formula DB 900 match at least from the beginning of the remaining character string after deletion by the deletion unit 1114. Then, the deleting unit 1114 deletes the first character from the remaining character string that has not been searched from the partial characteristic formula DB 900 by the dividing unit 1109. In the example of FIG. 14, the processing of the dividing unit 1109 and the deleting unit 1114 is repeated, so that there are no characters in the remaining character strings.

  Then, when there is no character to be deleted from the remaining character string, the second registration unit 1106 is searched by the search unit 1110 from the character string arranged in the character order deleted by the deletion unit 1114 and the extracted characteristic formula. A pair of the characteristic expression excluding the partial characteristic expression of the partial name is registered in the second database. For example, the second registration unit 1106 includes a character string “thiol” arranged in the order of deleted characters, a characteristic expression “SH” obtained by removing the partial characteristic expression “CH3” from the extracted characteristic expression “CH3SH”, and Are registered in the partial characteristic formula DB 900.

  As a result, the generation apparatus 1100 can increase the types of the characteristic expressions as elements to the partial characteristic expression DB 900 in order to reproduce the substance names and the characteristic expressions from the correspondence DB 200. Therefore, it is possible to improve the possibility that the generation apparatus 1100 can reproduce the substance name from the correspondence relation DB 200 and its expression formula.

<Correspondence Relationship DB 200 and Partial Indication Formula DB 900 Creation Processing Procedure>
FIG. 15 is a flowchart illustrating a creation processing procedure of the correspondence DB 200 and the partial characteristic formula DB 900 by the generation device 1100. First, the generation device 1100 determines whether there is unselected document information from the document information DB 100 (step S1501). When there is unselected document information from the document information DB 100 (step S1501: Yes), the generation apparatus 1100 selects unselected document information from the document information DB 100 (step S1502). Then, the generation device 1100 extracts a character string that satisfies a condition that represents a combination of a substance name and a sexual expression from the selected document information (step S1503).

  Next, the generating apparatus 1100 identifies the substance name and the characteristic expression from the extracted character string (step S1504), and registers the combination of the substance name and the characteristic expression in the correspondence DB 200 (step S1505). For example, the generating apparatus 1100 extracts “ethane (H3CH3)” that satisfies the synonymous expression from the document information, identifies “ethane” that is katakana as a substance name, and includes “CH3CH3” that includes the alphabet in parentheses. Identified as a formula. Then, for example, the generation device 1100 registers the set of the specified substance name “ethane” and the differential expression “CH3CH3” in the correspondence DB 200 as a new record.

  Then, the generation apparatus 1100 executes a characteristic expression componentization process on the new set of substance name and characteristic expression (step S1506), and returns to step S1501.

  FIG. 16 is a flowchart showing a detailed description of the characteristic type componentization processing (step S1506) shown in FIG. First, the generation device 1100 reads a newly registered partial name and partial characteristic formula from the correspondence DB 200 (step S1601). The characteristic type componentization rule application process is executed (step S1602). The characteristic expression componentization rule application process (step S1602) is a process of generating a combination of a partial name and a partial characteristic expression by applying the rule of the rule DB 400 to the combination of the substance name and the characteristic expression. is there. Details of the characteristic type componentization rule application processing (step S1602) will be described later.

  Thereafter, the generation apparatus 1100 registers the combination of the partial name and the partial characteristic formula obtained in the characteristic formula componentization rule application process (step S1602) in the partial characteristic formula DB 900 (step S1603). Then, the generation device 1100 determines whether or not the registration is a new registration for the combination of the partial name and the partial characteristic formula (step S1604). If it is a new registration (step S1604: Yes), the generation device 1100 determines whether or not the registration is applicable to a rule for a combination of a partial name and a partial characteristic formula (step S1605). For example, when the number of applied rules exceeds a preset upper limit, when there are no unapplied rules, or when the number of partial partial formulas generated exceeds a preset upper limit, For example, when a predetermined time has elapsed since the start of the conversion process (step S1506).

  If it is applicable (step S1605: Yes), the generation apparatus 1000 executes a characteristic type componentization process (step S1606). The characteristic type componentization process (step S1606) is the same process as the characteristic type componentization process (step S1506). As a result, the generation apparatus 1100 can recursively generate a pair of a partial name and a partial characteristic expression. Since the characteristic formula componentization process is recursively executed, in the case of Yes in step S1605, the process returns to step S1601. In step S1601, when there are a plurality of pairs of newly registered partial names and their partial characteristic expressions, the generation apparatus 1100 reads each of them, and performs a characteristic expression componentization rule application process (step S1602). Returning to, the recursive process is performed.

  In step S1604, if it is not a new registration (step S1604: No), since it has already been registered in the partial characteristic formula DB 900, the generating apparatus 1100 ends the characteristic formula componentization process and displays the characteristic formula componentization process. Return to the step that called. Also, when it is determined in step S1604 that application is not possible (step S1604: No), the generation device 1100 ends the characteristic type componentization process and returns to the step that called the characteristic type componentization process. In this manner, the combination of the partial name and the partial characteristic formula is recursively generated and registered in the partial characteristic formula DB 900 by the characteristic formula componentization processing (step S1506).

  FIG. 17 is a flowchart showing a detailed description of the characteristic expression componentization rule application processing (step S1602) shown in FIG. First, the generation device 1100 determines whether there is a rule in the rule DB 400 that matches the application condition (step S1701). If there is a rule that matches the application condition (step S1701: Yes), the generation device 1100 selects one of the rules that match the application condition (step S1702). Next, the generating apparatus 1100 changes the characteristic formula according to the selection rule (step S1703). The same applies to the case where the target expression is a partial expression.

  Then, the generation device 1100 determines whether or not the substance name is changed in the selection rule (step S1704). If not defined (step S1704: NO), the process returns to step S1701. On the other hand, when defined (step S1704: Yes), the generation device 1100 determines whether the defined change is an addition of a ending (step S1705). When the ending is added (step S1705: Yes), the generation device 1100 identifies the stem of the substance name from the stem dictionary 800 (step S1706), and proceeds to step S1707.

  On the other hand, if the ending is not added in step S1705 (step S1705: No), the generating apparatus 1100 moves to step S1707 because it adds an ending. In step S1707, the generation device 1100 generates a partial name by adding the ending or the beginning (step S1707), and the process returns to step S1701. In step S1701, when there is no rule that matches the application condition (step S1701: No), the generation device 1100 determines the part name obtained in the characteristic expression componentization rule application process (step S1602) and the partial characteristic expression. The set is registered in the partial characteristic formula DB 900 (step S1603). Note that the processing in steps S1704 to S1707 is the same when the target substance name is a partial name. In this way, the part name and the partial characteristic expression are generated by the characteristic expression componentization rule application process (step S1602).

<Confidence update processing procedure>
FIG. 18 is a flowchart showing a certainty factor update processing procedure by the generation device 1100. First, the generation device 1100 determines whether or not there is a pair of an unselected substance name and a characteristic formula from the correspondence DB 200 (step S1801). If there is one (step S1801: Yes), the generation device 1100 reads a pair of the unselected substance name and the characteristic formula from the correspondence DB 200 (step S1802). Here, the read substance name is referred to as an extracted substance name, and the read characteristic expression is referred to as an extracted characteristic expression. Then, the generation device 1100 executes the extracted substance name division process (step S1803). The extracted substance name dividing process will be described later with reference to FIG.

  The generation device 1100 determines whether there is an unselected divided part name among the plurality of divided part names (step S1804). If there is an unselected divided part name (step S1804: Yes), the generation device 1100 selects one selected divided part name (step S1805), and sets the partial characteristic expression of the selected divided part name to the partial characteristic. A search is performed from the formula DB 900 (step S1806), and the process returns to step S1804.

  On the other hand, in step S1804, when there is no unselected divided part name (step S1804: No), the generation device 1100 generates a connected characteristic expression by concatenating the partial characteristic expressions (step S1807), and the extraction characteristic It is determined whether or not the expression and the connection characteristic expression match (step S1808). When the extracted characteristic formula and the connection characteristic expression match (step S1808: Yes), the generation device 1100 updates the certainty factor of the partial characteristic expression constituting the connection characteristic expression in the partial characteristic expression DB 900 ( Step S1809) and the process proceeds to Step S1801. When the extracted characteristic formula and the connection characteristic formula do not match (step S1808: No), the generation device 1100 returns to step S1801.

  In step S1801, when there is no combination of an unselected substance name and a specific expression from the correspondence DB 200 (step S1801: No), the generation device 1100 ends a series of processes.

  FIG. 19 is a flowchart showing a detailed description of the extracted substance name division process (step S1804) shown in FIG. First, the generation device 1100 determines whether or not there is an unselected partial name in the partial characteristic formula DB 900 (step S1901). When there is an unselected partial name in the partial characteristic formula DB 900 (step S1901: Yes), the generation device 1100 selects one unselected partial name from the partial characteristic formula DB 900 (step S1902). The selected part name is referred to as “selected part name”.

  The generation apparatus 1100 determines whether the selected part name is included in the extracted substance name (step S1903). For example, the generation device 1100 determines by performing a longest match search from the beginning of the extracted substance name. For example, when the selected portion name is “ethane” and the extracted substance name is “ethanesulfonic acid”, the longest matching selected portion name “ethane” is included in the extracted substance name “ethanesulfonic acid”.

  When the selected part name is not included in the extracted substance name (step S1903: No), the generation apparatus 1100 returns to step S1901 to newly select the part name. On the other hand, when the selected part name is included in the extracted substance name (step S1903: Yes), the generation device 1100 holds the selected part name as a divided part name from the extracted substance name (step S1904). In the above case, the selected part name “ethane” is held as the divided part name.

  After that, the generation apparatus 1100 newly sets the remaining character string excluding the selected part name as the extracted substance name (step S1905). For example, in the above-described case, the generation apparatus 1100 newly sets the residual character string “sulfonic acid” obtained by removing the selected partial name “ethane” from the extracted substance name “ethanesulfonic acid” as the extracted substance name. Then, the process returns to step S1901.

  In this way, steps S1901 to S1905 are repeated until there are no extracted substance names. In the case described above, when “sulfonic acid” appears as a selected part name for a new extracted substance name “sulfonic acid”, the selected part name “sulfonic acid” is included in the extracted substance name “sulfonic acid”. Therefore, the generation device 1100 holds the selected part name “sulfonic acid” as the divided part name. On the other hand, when “sulfonic acid” is not registered in the partial characteristic formula DB 900, the extracted substance name “sulfonic acid” is not held as a divided part name.

  Thereafter, when there is no unselected partial name in step S1901 (step S1901: No), the generation device 1100 determines whether the division is successful (step S1906). For example, if there is no extracted substance name that is not held as a divided part name, a plurality of divided part names are held, and the division is successful. In the above-described example, when “sulfonic acid” is held as the divided part name, the name of the extracted substance does not remain, and thus the generation apparatus 1100 determines that the division is successful. On the other hand, when “sulfonic acid” is not retained as the name of the divided part, “sulfonic acid” remains as the name of the extracted substance, and thus the generation apparatus 1100 determines that the division has failed.

  If the division is successful (step S1906: YES), the generation device 1100 associates and stores the held division part name (step S1907). Then, the process proceeds to step S1804, and the associated divided part name becomes a selection target in step S1805. On the other hand, in the case of division failure (step S1906: No), since the extraction substance name in step S1904 cannot be divided, the generation apparatus 1100 executes registration processing (step S1908) and returns to step S1801. In the above-described example, the registration process (step S1908) is a process of newly registering “sulfonic acid” that has not been held as the divided part name in the partial characteristic formula DB 900. The registration process will be described in detail with reference to FIG. In this way, in the extracted substance name dividing process (step S1803), the generation device 1100 can divide the character string of the substance name into a plurality of partial names. Furthermore, when the division of the substance name character string fails, a new record is registered in the partial referential expression DB 900.

  FIG. 20 is a flowchart showing a detailed description of the registration process (step S1908) shown in FIG. First, the generation apparatus 1100 determines whether there is a divided part name (step S2001). When there is no divided part name (step S2001: No), the generation apparatus 1100 registers a set of the extracted characteristic formula and the extracted substance name in the partial characteristic formula DB 900 (step S2007), and the process returns to step S1801. As a result, a new set of the formula and the substance name is registered in the partial formula DB 900.

  If there is a divided part name (step S2001: Yes), the generation device 1100 determines whether there is an unselected divided part name among the plurality of divided part names (step S2002). When there is an unselected divided part name among a plurality of divided part names (step S2002: Yes), the generation device 1100 selects one selected divided part name (step S2003), and displays the partial divided part name. The sex formula is searched from the partial characteristic formula DB 900 (step S2004), and the process returns to step S2002.

  On the other hand, when there is no unselected division part name among the plurality of division part names (step S2002: No), the generation device 1100 generates a linked characteristic expression by concatenating the partial characteristic expressions (step S2005). ). The generation apparatus 1100 registers a pair of the remaining characteristic expression obtained by removing the connected characteristic expression from the extracted characteristic expression and the divided part name in the partial characteristic expression DB 900 (step S2006), and returns to step S1801.

<Registration Processing Procedure of Partial Indication Formula DB900>
FIG. 21 is a flowchart illustrating an example of a registration processing procedure of the partial referential expression DB 900 by the generation device 1100. The generation apparatus 1100 determines whether or not there is a combination of an unselected substance name and a characteristic formula from the correspondence DB 200 (step S2101). When there is a pair of an unselected substance name and an explicit expression (step S2101: Yes), the generation apparatus 1100 reads an unselected substance name and an explicit expression pair from the correspondence DB 200 (step S2102). Here, the read substance name is referred to as an extracted substance name, and the read characteristic expression is referred to as an extracted characteristic expression.

  Next, the generation device 1100 determines whether or not there is an unselected partial name in the partial referential expression DB 900 (step S2103). When there is an unselected partial name (step S2103: Yes), the generation device 1100 selects one unselected partial name from the partial referential expression DB 900 (step S2104). The generation apparatus 1100 determines whether the selected part name is included in the extracted substance name (step S2105).

  When the selected part name is not included in the extracted substance name (step S2105: No), the process returns to step S2103. On the other hand, when the selected part name is included in the extracted substance name (step S2105: Yes), the generating apparatus 1100 moves to step S2106, and the generating apparatus 1100 extracts the remaining character string excluding the selected part name as the extracted substance. The name is set (step S2106). For example, when the selected part name is “methane” and the extracted substance name is “methanethiol”, the longest matching selected part name “methane” is included in the extracted substance name “methanethiol”. The generation apparatus 1100 newly sets a residual character string “thiol” obtained by removing the selected part name “methane” from the extracted substance name “methanethiol” as the extracted substance name.

  Then, the generation apparatus 1100 searches the partial characteristic formula DB 900 for the selected partial name from the partial characteristic formula DB 900 (step S2107), and sets the residual characteristic expression excluding the extracted characteristic expression as the extracted characteristic (step S2107). S2108). In the above case, the generation apparatus 1100 searches for the partial characteristic formula “CH3” of the selected partial name “methane”. The generation apparatus 1100 newly sets the residual characteristic expression “SH” obtained by removing the partial characteristic expression “CH3” from the extraction characteristic expression “CH3SH” as the extraction characteristic expression.

  In step S2103, when there is no unselected partial name (step S2103: No), the generation apparatus 1100 deletes the first character from the extracted substance name (step S2109), and the generation apparatus 1100 displays the deletion character. The deletion character string is generated by concatenating (step S2110). In the above case, for example, the generation device 1100 deletes the first character “Chi” from the extracted substance name “thiol” and generates a deleted character string “Chi”. Then, the generation apparatus 1100 repeats the processing of step S2103 to step S2108 for the extracted substance name “all” after deletion.

  The generation apparatus 1100 determines whether the extracted substance name is empty (step S2111). If the extracted substance name is not empty (step S2111: NO), the process returns to step S2103. On the other hand, when the extracted substance name is empty (step S2111: Yes), the generation device 1100 registers the combination of the deleted character string and the extracted characteristic formula in the partial characteristic formula DB 900 (step S2112), and the process proceeds to step S2101. Return. In the above case, since there is no partial name in the partial referential expression DB 900 that matches the extracted substance names “all”, “all”, and “le”, the generation device 1100 deletes all the characters in the extracted substance name. End up. Therefore, the generation device 1100 registers a set of the deletion character string “thiol” and the extracted characteristic expression “SH” in the partial characteristic expression DB 900.

  In step S2101, if there is no unselected substance name / indicative formula pair (step S2101: No), the series of processing ends. Further, the generation device 1100 alternately executes the process of removing the reliability failure process shown in FIG. 18 except for the case of the division failure and the registration process of the partial characteristic formula DB 900 shown in FIG. Also good.

  As described above, the generation apparatus, generation program, and generation method according to the present embodiment extract a part that is presumed that a substance name and a chemical formula are described in synonymous expressions from document information. , Accumulate substance name and chemical formula pairs in the database. Thus, the generation apparatus can automatically find a candidate for a combination of a substance name and its chemical formula from electronic information. In addition, although a dictionary showing the correspondence between the names of substances sold and their chemical formulas is expensive, this generator can automatically create a database from existing chemical and pharmaceutical electronic documents. Costs for creating a new chemical formula from the chemical formula can be reduced.

  In addition, the generation device applies the chemical reaction rule to the chemical formula of the substance name included in the first database created from the electronic document, and registers it in the second database. Thereby, the production | generation apparatus can improve the production | generation precision of the chemical formula derived from chemical formula, and can produce a 2nd database automatically.

  Moreover, the set of the substance name and its chemical formula included in the first database is not always accurate. Therefore, the combination of the substance name and chemical formula in the second database obtained by applying the chemical reaction rule to the combination of the substance name and chemical formula in the first database is not always accurate. Therefore, since the generation device has a plurality of combinations of substance names and chemical formulas registered in the second database, the generation device connects the substance names and chemical formula combinations in the second database, and It is determined whether or not a set of a substance name and its chemical formula in one database can be reproduced. Then, the generation apparatus determines that the combination of the substance name and the chemical formula in the second database that can reproduce the combination of the substance name and the chemical formula in the first database has high accuracy. Therefore, the user can use this accuracy as an indicator of which chemical formulas in the second database are useful and may exist.

  In addition, if the number of registered types of combinations of substance names and chemical formulas registered in the second database is small, the substance names and chemical formulas in the first database may not be reproduced. Therefore, if the partial name of the substance name extracted from the electronic document and the partial explicit expression of the explicit name expression of the substance name are not registered in the second database, the generating device can detect the partial name and the partial name. Register a new set of indicia expressions in the database. As a result, the generation device can increase the number of types registered in the database. The generation device alternately repeats the accuracy determination process in the second database and the process of increasing the number of registered types in the second database, so that a useful indication can be obtained from the characteristic expression registered in the second database. You can find the sex formula.

  Note that the generation method described in the present embodiment can be realized by executing a generation program prepared in advance on a computer such as a personal computer or a workstation. The generation program is recorded on a computer-readable recording medium such as a hard disk, a flexible disk, a CD-ROM, an MO, and a DVD, and is executed by being read from the recording medium by the computer. The generation program may be distributed via a network such as the Internet.

  The following additional notes are disclosed with respect to the embodiment described above.

(Supplementary note 1) A character string extraction unit that extracts a character string satisfying a condition representing a combination of a substance name and a chemical formula of the substance name from document information;
A first registration unit that registers a set of the substance name and the chemical formula included in the character string extracted by the character string extraction unit in a first database;
A generation apparatus comprising:

(Supplementary Note 2) A rule extraction unit that extracts one of the rules from the rule group that defines the condition related to the chemical reaction to be applied and the conversion content when the condition is satisfied,
A determination unit that determines whether the chemical formula registered in the first database by the registration unit satisfies a condition to be applied in any of the rules extracted by the rule extraction unit;
A conversion unit that converts the substance name and the chemical formula, which are a set of the chemical formulas, according to the conversion content when it is determined by the determination unit to satisfy;
A second registration unit for registering the chemical formula and substance name converted by the conversion unit in a second database;
The generating apparatus according to appendix 1, wherein:

(Supplementary Note 3) The second database has a certainty factor representing the certainty of the set of the converted chemical formula and the converted substance name for each set of the converted chemical formula and the converted substance name. And
When the extracted substance name included in the first database includes the first substance name in the second database, the registered substance name is determined based on the first substance name and the extracted substance name from the first substance name. A second substance name excluding the substance name of
The first chemical formula of the first substance name divided by the dividing unit is searched from the second database, and the second chemical formula of the second substance name divided by the dividing unit is searched for the second chemical formula. A search unit for searching from a database;
A generating unit that generates a linked chemical formula obtained by connecting the first chemical formula and the second chemical formula searched by the search unit;
A determination unit that determines whether or not the linked chemical formula generated by the generation unit matches the extracted chemical formula that is a set of the extracted substance names included in the first database;
When the determination unit determines that they match, the certainty factor representing the certainty of the set of the first substance name and the first chemical formula included in the second database, the second substance name, and the second An update unit that updates a certainty factor representing the certainty of the set of chemical formulas to a value indicating more certainty,
The generating apparatus according to appendix 2, characterized by comprising:

(Appendix 4) The update unit
When it is determined by the determination unit that they match, the certainty factor representing the certainty of the set of the first substance name and the first chemical formula and the certainty of the set of the second substance name and the second chemical formula The generating apparatus according to supplementary note 3, wherein the certainty factor representing the above is higher than that before update.

(Supplementary Note 5) The search unit
If the second substance name is not included in the second database, the first chemical formula is searched from the second database,
The second registration unit includes:
When the second substance name is not included in the second database, the first chemical substance searched by the search unit among the extracted chemical formulas that are a set of the extracted substance names included in the first database. The generating apparatus according to claim 3, wherein a set of the chemical formula excluding the chemical formula and the second substance name is registered in the second database.

(Additional remark 6) When the 2nd substance name is not contained in the 2nd database, it has a deletion part which deletes the first character from the 2nd substance name,
The deletion unit is
When the second substance name after deletion by the deletion unit is not included in the second database, the first character is further deleted from the second substance name,
The second registration unit includes:
When there is no character deleted from the second substance name by the deletion unit, a character string arranged in the order of characters deleted by the deletion unit, and a chemical formula obtained by removing the first chemical formula from the extracted chemical formula The generating apparatus according to appendix 5, wherein a set is registered in the second database.

(Appendix 7) The computer
From the document information, extract a character string that satisfies the condition representing the combination of the substance name and the chemical formula of the substance name,
Registering the set of the substance name and the chemical formula contained in the extracted character string in the first database;
A generation method characterized by executing processing.

(Appendix 8)
From the document information, extract a character string that satisfies the condition representing the combination of the substance name and the chemical formula of the substance name,
Registering the set of the substance name and the chemical formula contained in the extracted character string in the first database;
A generation program characterized by causing processing to be executed.

200 Correspondence DB
900 Partial Indication Formula DB
100-1 to 100-n Document information 1101 Character string extraction unit 1102 First registration unit 1103 Rule extraction unit 1104 Judgment unit 1105 Conversion unit 1106 Second registration unit 1109 Division unit 1110 Search unit 1111 Generation unit 1112 Determination unit 1113 Update Part 1114 Deletion part

Claims (6)

A character string extraction unit that extracts a character string satisfying a condition representing a set of a substance name and a chemical formula of the substance name from document information;
A first registration unit that registers a set of the substance name and the chemical formula included in the character string extracted by the character string extraction unit in a first database;
A generation apparatus comprising: A rule extraction unit that extracts one of the rules from the rule group that defines the conditions related to the chemical reaction to be applied and the conversion contents when the conditions are satisfied;
A determination unit that determines whether the chemical formula registered in the first database by the registration unit satisfies a condition to be applied in any of the rules extracted by the rule extraction unit;
A conversion unit that converts the substance name and the chemical formula, which are a set of the chemical formulas, according to the conversion content when it is determined by the determination unit to satisfy;
A second registration unit for registering the chemical formula and substance name converted by the conversion unit in a second database;
The generating apparatus according to claim 1, wherein: The second database has a certainty factor representing the certainty of the set of the converted chemical formula and the converted substance name for each set of the converted chemical formula and the converted substance name,
When the extracted substance name included in the first database includes the first substance name in the second database, the extracted substance name is determined from the first substance name and the extracted substance name. A second substance name excluding the substance name of
The first chemical formula of the first substance name divided by the dividing unit is searched from the second database, and the second chemical formula of the second substance name divided by the dividing unit is searched for the second chemical formula. A search unit for searching from a database;
A generating unit that generates a linked chemical formula obtained by connecting the first chemical formula and the second chemical formula searched by the search unit;
A determination unit that determines whether or not the linked chemical formula generated by the generation unit matches the extracted chemical formula that is a set of the extracted substance names included in the first database;
When the determination unit determines that they match, the certainty factor representing the certainty of the set of the first substance name and the first chemical formula included in the second database, the second substance name, and the second An update unit that updates a certainty factor representing the certainty of the set of chemical formulas to a value indicating more certainty,
The generating device according to claim 2, wherein The search unit
If the second substance name is not included in the second database, the first chemical formula is searched from the second database,
The second registration unit includes:
When the second substance name is not included in the second database, the first chemical substance searched by the search unit among the extracted chemical formulas that are a set of the extracted substance names included in the first database. The generating apparatus according to claim 3, wherein a set of a chemical formula excluding the chemical formula and the second substance name is registered in the second database. Computer
From the document information, extract a character string that satisfies the condition representing the combination of the substance name and the chemical formula of the substance name,
Registering the set of the substance name and the chemical formula contained in the extracted character string in the first database;
A generation method characterized by executing processing. On the computer,
From the document information, extract a character string that satisfies the condition representing the combination of the substance name and the chemical formula of the substance name,
Registering the set of the substance name and the chemical formula contained in the extracted character string in the first database;
A generation program characterized by causing processing to be executed.

Images