JPH10312380A - E-mail subject generation method and apparatus, and storage medium storing e-mail subject generation program - Google Patents

Abstract

(57) [Summary] [Problem] E-mail subject generation that automatically generates subjects for e-mails with no restrictions on input and reads out automatically generated subjects even for e-mails that do not have a subject described Provided are a method and apparatus and a storage medium storing an email subject generation program. The present invention is directed to a pattern / dictionary having a collation pattern composed of a word string pattern that can be a subject by decomposing the body of an e-mail into word strings, and a generation pattern composed of a word string pattern that generates the subject. , The decomposed word string is collated with the matching pattern, and a subject is generated from the collated word string.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for generating an e-mail subject and a storage medium storing an e-mail subject generation program. More particularly, the present invention relates to an email subject generation method and apparatus for generating a subject of header information, and a storage medium storing an email subject generation program.

[0002]

2. Description of the Related Art E-mail is sent to RFC 822 (RFC: Reque
st for Comments, TCP / IP conventions), stipulates that it consists of a header and a body. In the header, fields such as a subject, a sender, a receiver, and a transmission date and time are described as delivery information of the electronic mail. this house,
The subject describes the subject of the e-mail in order to improve the overview of the received e-mail.

As an example of automatically setting a subject from the body of a mail, there is Japanese Patent Application No. Hei 6-198113 “Mail system having automatic mail subject setting function”. This system is a method of automatically setting the first line of a mail as a subject, and the sender can use the mail by closing the subject of the mail in a format described at the beginning of the mail.

[0004]

However, the above conventional system is not effectively used for subject extraction for the following reasons. Mailers (tools for reading and writing e-mails) are developed mainly in the United States and may not support Japanese. Such mailers have a problem that subjects can only be entered in alphabetical characters when sending, and Japanese subjects cannot be read when receiving.There are many mails that do not have a subject or mail that has a subject entered in Rome. doing.

[0005] In addition, even when a Japanese-language mailer is used, it is often not input because of the trouble of inputting the subject. This is data that the subject is primarily intended to improve the convenience of the recipient,
This is because there is little merit for the sender. Therefore,
The recipient of the e-mail needs to read the information including the information unnecessary for him / her, which is inconvenient in that it is troublesome to search only the information that is really necessary.

Furthermore, the method disclosed in Japanese Patent Application No. Hei 6-198113 is a method in which the first line of a mail is automatically set as a subject, and the sender describes the subject of the mail at the beginning. The format needs to be limited. In general, the beginning of an e-mail is various, and a recipient name, a greeting, and the like are described according to the preference of the sender. There is a problem that a subject is erroneously generated for such an email.

[0007] The present invention has been made in view of the above points, and stores an electronic mail subject generation method and apparatus and an electronic mail subject generation program for automatically generating a subject for a mail having no restriction on input. It is intended to provide a storage medium. Another object of the present invention is to read out an automatically generated subject even for an email in which a subject is not described in an electronic mail access system in which an electronic mail is converted into a voice and can be listened to by telephone at a place where the user is away. Accordingly, the present invention provides a method and apparatus for generating an e-mail subject and a storage medium storing an e-mail subject generation program, which enable a user to detect the contents of an e-mail arriving in a short time without listening to and judging the entire text. That is.

[0008]

FIG. 1 is a diagram for explaining the principle of the present invention. According to the present invention, in an electronic mail subject generation method for generating an electronic mail subject, the body of the electronic mail is decomposed into word strings (step 1), and the decomposed word strings are collated with a matching pattern (step 2). Then, a subject is generated from the collated word string (step 3).

According to the present invention, when the body of the e-mail is decomposed into a word string in step 1 described above, the part of speech of a word constituting the body of the e-mail is determined by morphological analysis processing, and a word example is determined. Generate. Further, in the present invention, when matching the decomposed word string with the matching pattern in step 2, a matching pattern including a word string pattern that can be a subject and a generation pattern including a pattern of a word string that generates the subject Is referenced.

Further, according to the present invention, when collating the decomposed word string with the collation pattern in step 2, the collation is performed based on the position where the word string is collated with the collation pattern or the collation condition defining the word attribute. Collate. Further, in the present invention, when generating a subject in step 3, the subject is generated by referring to a part-of-speech conversion table for changing the part of speech of the word string of the body of the e-mail to the part of speech of the generation pattern.

Further, according to the present invention, the generated subject is converted into voice and output as it is or to a voice terminal via a network. FIG. 2 is a diagram illustrating the principle of the present invention. The present invention relates to an e-mail subject generating apparatus for generating an e-mail subject, comprising: a word string decomposing means for decomposing the body of an e-mail into word strings; and a word string decomposed by the word string decomposing means. And a subject generating unit 40 for generating a subject from the word string collated by the collating unit 30.

The word string decomposing means 10 determines a part of speech of a headline of a word constituting the body of the e-mail,
It includes morphological analysis means for generating a word example. Further, the word string decomposing means 10 includes a kana converting means for converting a Roman character into a kana or a hiragana when the character string constituting the body of the e-mail is a Roman character.

Further, the word string decomposing means 10 includes a foreign language word string decomposing means for decomposing the character string for each word when the character string constituting the body of the e-mail is a foreign language. Also, the electronic mail subject generating apparatus of the present invention has a pattern pair dictionary having a matching pattern consisting of a word string pattern that can be a subject and a generating pattern consisting of a pattern of a word string that generates the subject. Includes means for referring to a pattern pair dictionary.

Further, the pattern pair dictionary further has a collation condition for defining a word string and a collation pattern or a word attribute, and the collation means 30 performs collation based on the collation condition. including. Further, the e-mail subject generation device of the present invention further has a part-of-speech conversion table for changing the part of speech of the word string of the body of the e-mail to the part of speech of the generation pattern, and the subject generation means refers to the part-of-speech conversion table, Means for generating the subject.

Further, the electronic mail subject generation apparatus of the present invention provides a foreign language pattern having a collation pattern consisting of a foreign language word string pattern that can be a subject and a generation pattern consisting of a foreign language word string pattern that produces the subject. When a word in a foreign language is acquired by the word string decomposing means 10, the matching means 30 includes means for referring to the foreign language pattern pair dictionary.

Further, the electronic mail subject generating apparatus of the present invention comprises a voice converting means for voice-converting the subject generated by the subject generating means, and a voice converted by the voice converting means as it is or via a network. It further has audio output means for outputting to the terminal. The storage medium storing the e-mail subject generation program of the present invention includes: an e-mail acquisition process for inputting the text of the e-mail; a word sequence decomposition process for decomposing the e-mail into a word sequence; and a word sequence that can be a subject. A matching process of referring to a pattern pair dictionary having a matching pattern consisting of a pattern and a generation pattern consisting of a pattern of a word string for generating the subject, and matching the matching pattern with the word string decomposed in the word string decomposing process. And a subject generation process of generating a subject by referring to a part-of-speech conversion table that changes the part of speech of the word string of the body of the e-mail to the part of speech of the generation pattern based on the word string collated by the collation process.

[0017] Further, there is further provided a voice conversion process for voice-converting the subject obtained by the subject generation process, and a voice output process for voice-outputting the voice-converted subject when a voice output request is received. As described above, in the present invention, when the subject is not described in the received e-mail, the body of the e-mail is divided into words, and the words obtained by collating the divided words with a predetermined collation pattern are used. It is possible to generate a subject based on the subject. At this time, it is possible to generate a subject not only in a normal Japanese sentence but also in a romanized electronic mail body by converting the romanized character description into a normal Japanese sentence.

Further, when the body of the e-mail is in a foreign language such as English, words are decomposed in accordance with the grammar of the foreign language and the foreign language pattern pair dictionary to be applied to the foreign language is referred to. In the same manner as described above, the collation processing can be performed, and the subject can be generated based on the collated words, so that there is no restriction on the e-mail to be received.

Further, when the user issues a read request for the subject from a voice terminal such as a telephone, the generated subject is voice-converted to a network (telephone network).
The user can listen to the automatically generated subject by voice even when receiving an e-mail in which the subject is not described.

Further, the above-described operation is constructed as a program, stored in a portable medium, and distributed to a user. If the program is installed on the user's personal computer, the subject is described in the received e-mail. Even if there is no subject, a subject is automatically generated and output.

[0021]

FIG. 3 shows the configuration of an electronic mail subject generation apparatus according to the present invention. The e-mail subject generation device shown in FIG. 1 includes a morphological analysis unit 10, a pattern pair dictionary 20, a subject collation unit 30, and a subject generation unit 40. The e-mail is input, and the generated subject is output.

The morphological analyzer 10 inputs an electronic mail, divides the text into words, and determines the part of speech.
As a result, a body word string that holds the headline and part of speech of each word that forms the body is generated. Such processing is called morphological analysis, and has already been established as a technique that can withstand practical use. The pattern pair dictionary 20 holds various kinds of information for extracting an expression that can be a subject and generating a subject.

The subject matching unit 30 matches the word string generated by the morphological analysis unit 10 with a matching pattern of an expression that can be a subject, and outputs a matched matching pattern. The subject generation unit 40 generates a subject from a generation pattern corresponding to this collation pattern on a one-to-one basis.

FIG. 4 shows an example of the configuration of a pattern pair dictionary according to the present invention. In the figure, a matching pattern 100 of the pattern pair dictionary 20 is a set of word strings for extracting an expression serving as a subject material from a mail body. Each box in the figure represents a word, the notation of which is shown below. Matches when the body word string includes the word strings of these notations. The positions of the word strings are referred to as a first row, a second row,... From the top.
The positions of the words in the word sequence are referred to as a first column, a second column, ... from the left.

First, the word notation of the matching pattern will be described. (A) Notation = [(part of speech)] When there is a square bracket in the notation, the characters in it indicate the part of speech. (B) Notation = * Matches any word sequence. That is, in the case where there is no word, in the case of one word, and in the case of a plurality of words, matching is performed.

(C) Notation = (heading) When there is a character other than the above in the notation, it indicates the heading of the word. For example, the matching patterns “[noun]”, “ni”, “tsu” and “te” on the first line are “words whose part of speech is noun” “words whose heading is“ ni ”” “headings are“ tsu ” Word "and" word whose heading is "te"". The matching patterns "[noun]", "wo", "*" and "[sa-variant noun]" in the third line are "words whose part of speech is noun""words whose heading is" wo """words of any word Line "and" a word whose part of speech is a paranoun ".

Here, "*" matches a sequence of words not related to the subject and is excluded in subject generation. Generated pattern 110 of pattern versus dictionary 20
Is a set of word strings of the expression to be the subject. The definition of the word notation of the generated pattern is basically the same as the matching pattern 100. However, the characters in parentheses are (numbers):
(Part of speech). This means that the word in the (numerical) column of the word string of the matching pattern 100 including the notation is (part of speech)
To be output as a subject.
At this time, if the body word string matches the word string of the matching pattern 100, the subject generation unit 40 replaces the text string with the word string of the generation pattern 110 on the same line. At this time, basically, words that match the matching pattern are output as they are, but in some cases, the part of speech may be changed between the matching pattern and the generated pattern. For example, it is more common to express a subject with a noun phrase than with a sentence.
Convert to Here, the sa-variant noun is a noun that is made into a verb by connecting "su" to the back.

The subject generation unit 40 collates the body word string generated by the morphological analysis unit 10 using the matching pattern 100, and replaces the word string of the matched matching pattern 100 with the word string of the generated pattern 110 on the same line. I do. This replacement method is as follows. Here, the matching pattern 100 and the generated pattern 11 associated with each other by a numeral
The word of 0 is called a source word and a target word, respectively.

For example, the word in the fourth column of the fourth row of the matching pattern 100 and the word in the third column of the fourth row of the generated pattern are a source word and a target word, respectively. The method of generating the subject (condition) is shown below. (1) When the notation of the target word includes parentheses, that is, when notation = [(number) :( part of speech)]: (1-1) When the part of speech of the source word = (part of speech); Output the words in the word string.

(1-2) When the part of speech of the source word is ≠ (part of speech): The words in the body word string that match the source word are replaced by the part of speech conversion table 120 and output. (2) When the target word notation does not include parentheses,
That is, when notation = (heading); the target word is output.

Part-of-speech conversion table 1 of pattern pair dictionary 20
20 records the word pairs before and after the replacement in order to replace the words in the body word string according to the generation condition (1-2). The word before replacement is in the left column, and the word after replacement is in the right column. FIG. 4 shows an example of a word pair in the case where the part of speech of the matching pattern 100 of the pattern pair dictionary 20 is a verb, and the part of speech of the generation pattern 110 is a paranoun.

The matching condition 130 of the pattern pair dictionary 20 is as follows.
A position condition and a word condition that respectively specify a matching range and a word attribute when matching the mail text with the matching pattern 100 are described. For example, as the position condition, the range to be collated is described as being within the first x sentences of the mail text (x: is a positive number).
This is because the expression including the subject of the mail that can be the subject is often written at the beginning. In addition, as a word condition, when the word of the matching pattern 100 is a noun, a descriptive noun (“above”, “this case”, etc.) and an abstract noun (“item”, “information”, etc.) are excluded from the matching target. This is because words generally used in various sentences are not suitable as subjects.

FIG. 5 is a flowchart showing the operation of the present invention. Step 101) The mail text is input to the morphological analysis unit 10. Step 102) The morphological analysis unit 10 divides the input mail text into text word strings. Step 103) The subject matching unit 30 performs matching with the matching pattern 100 while checking the matching condition 130 of the pattern pair dictionary 20.

Step 104) Subject generation unit 4
0 generates and outputs a subject according to the subject generation method.

[0035]

Embodiments of the present invention will be described below with reference to the drawings. The operation of the above-described electronic mail subject generation device will be described based on the operation of FIG. 5 based on the example of FIG. <Example 1> in FIG. 6 shows the processing result of the text “About the in-house tour”.

The morphological analysis unit 10 divides the mail body into a body word string and passes it to the subject matching unit 30. The subject matching unit 30 checks the matching condition 1 of the pattern pair dictionary 20.
While checking 30, the word string is collated with the collation pattern 100. As a result, it matches the word string in the first row of the matching pattern 100. Here, it is assumed that the position condition and the word condition of the matching condition 130 are satisfied. Generation pattern 11
The first column of the first row of 0 is a noun. Since this matches the part of speech of the source word, the subject generation unit 40 applies the above-described subject generation condition (1-1). As a result, the noun “in-house tour” is output as the subject.

<Example 2> of FIG. 6 shows the processing result of the text "Consultation for product announcement starts at 10:00". In the same procedure as in Example 1, subject matching unit 30 checks matching pattern 10
Matches the word string in the fifth row of 0. Here, “10:00” and “kara” in the body word string match “*” in the matching pattern. "10:00" as an unnecessary word for the subject
Delete "from". Now, it is assumed that the position condition and the word condition of the matching condition 130 are satisfied. The subject generation unit 40 performs the above-described subject generation conditions (1-1) in order from the first word of the word string on the fifth row of the generation pattern 110,
Applying (2), (1-1), (2), and (1-2),
As a subject, "product announcement""no""meeting"
"No" and "execute" are output. Here, the fifth column of the fifth row of the generation pattern is “sa-variable noun”, which does not match the part-of-speech “verb” of the source word.
, The verb “do” is converted to the noun “implement”.

Example 3 shows the processing result of the text "Please send improvement measures for this case". Subject collation unit 30
Matches the body word string “this case”, “ni”, “tsu”, “te” with the word string on the first line of the matching pattern 100. However, the noun “this case” in the body word string is an instruction noun, and the matching condition 13
Since the word condition of 0 is not satisfied, no match is made. next,
The subject matching unit 30 matches the “send improvement measure” in the body word string with the word string in the third row of the matching pattern 100. Here, it is assumed that the position condition of the matching condition 130 is satisfied. The subject generation unit 40 generates the generation pattern 1
The subject generation conditions (1-1), (2), and (1-1) are applied in order from the first word of the word string in the third row of No. 10, and "improvement measures", "no", and "send" are sent as subjects. "
Is output.

In the above-described embodiment, the input word string is described using an example in which Japanese characters are mixed with normal kanji characters. However, the present invention is not limited to this example, and the mail body is described in Roman characters. Even if it is performed, a romaji / kana conversion unit (not shown) is provided at the preceding stage of the morphological analysis unit 10, and the word string converted to kana (or hiragana) is output to the morphological analysis unit 10, A subject can be acquired in the same manner as in the case of ordinary Japanese as described above.

Further, when the body of the e-mail is described in a foreign language, a foreign language word division unit (not shown) is provided in place of the morphological analysis unit 10, and further, in place of the pattern pair dictionary 20. By providing a foreign language pattern pair dictionary, it is possible to obtain a subject by referring to the foreign language pattern dictionary with words that have been decomposed, as in the case of Japanese.

Further, the generated subject is converted into voice, and when the user issues a subject voice output request using a telephone, the subject voice is output via a network in response to the request. It becomes possible. In addition,
The present invention is not limited to the embodiments described above, and various changes can be made within the scope of the claims.

[0042]

As described above, according to the present invention, the subject is automatically generated in the transmission of the mail, thereby saving the trouble of inputting, and the subject is not input in the reception of the mail. Alternatively, by generating a Japanese subject for a mail whose subject is a Roman character, it is possible to improve the mail listability. Further, even when the subject is in English, the subject can be generated in the same manner as in Japanese by referring to the word decomposition or the foreign language pattern pair dictionary.

Also, in an electronic mail access system in which an electronic mail is vocalized and can be listened to by telephone at a place where the user is away, even if the electronic mail does not include a subject, the automatically generated subject is read out and used. This allows the user to detect the content of the arrived mail in a short time without having to listen to the entire text to make a decision.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the principle of the present invention.

FIG. 2 is a principle configuration diagram of the present invention.

FIG. 3 is a configuration diagram of an electronic mail subject generation device of the present invention.

FIG. 4 is a configuration example of a pattern pair dictionary of the present invention.

FIG. 5 is a flowchart showing the operation of the present invention.

FIG. 6 is a diagram for explaining the operation of one embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Word string decomposition | disassembly means, morphological analysis part 20 Pattern pair dictionary 30 Matching means, subject matching part 40 Subject generation means, subject generation part 100 Matching pattern 110 Generation pattern 120 Part-of-speech conversion table 130 Matching conditions

Claims (17)

[Claims] 1. An electronic mail subject generation method for generating an electronic mail subject, comprising the steps of: decomposing the body of the electronic mail into word strings; collating the decomposed word strings with a matching pattern; Generating an e-mail subject from an e-mail subject. 2. The method according to claim 1, wherein, when decomposing the body of the e-mail into a word string, a part of speech of a word constituting the body of the e-mail is determined by morphological analysis processing, and a word example is generated. Email subject generation method. 3. A pattern pair having a matching pattern consisting of a word string pattern that can be a subject and a generation pattern consisting of a pattern of a word string that generates the subject when matching the decomposed word string with a matching pattern. The method according to claim 1, wherein the dictionary is referred to. 4. The method according to claim 1, wherein when the decomposed word string is compared with a collation pattern, the collation is performed based on a position at which the word string is collated with the collation pattern or a collation condition defining a word attribute. Alternatively, the e-mail subject generation method described in 3. 5. The electronic device according to claim 1, wherein, when generating the subject, the subject is generated by referring to a part-of-speech conversion table that changes a part of speech of a word string in the body of the electronic mail to a part of speech of the generated pattern. Email subject generation method. 6. The electronic mail subject generation method according to claim 1, wherein the generated subject is voice-converted and output as it is or to a voice terminal via a network. 7. An e-mail subject generation device for generating an e-mail subject, comprising: a word string decomposing means for decomposing a body of the e-mail into a word string; and a word decomposed by the word string decomposing means. An e-mail subject generating apparatus, comprising: matching means for matching a column with a matching pattern; and subject generating means for generating a subject from the word string matched by the matching means. 8. The e-mail subject generation device according to claim 7, wherein the word string decomposition unit includes a morphological analysis unit that determines a part of speech of a word constituting the body of the e-mail and generates a word example. 9. The word string decomposing means according to claim 8, further comprising: if the character string constituting the body of the electronic mail is a Roman character, convert the Roman character to kana or hiragana. Email subject generation device. 10. The word string decomposing means includes a foreign language word string decomposing means for decomposing the character string for each word when a character string constituting the body of the e-mail is a foreign language. 8. The e-mail subject generation device according to 8. 11. A pattern pair dictionary having a matching pattern consisting of a word string pattern that can be a subject and a generation pattern consisting of a pattern of a word string that generates the subject, wherein the matching unit stores the pattern pair dictionary. 8. The electronic mail subject generation device according to claim 7, further comprising a reference unit. 12. The pattern pair dictionary further includes a collation condition for defining a position or a word attribute for collating the word string with the collation pattern, wherein the collation unit performs collation based on the collation condition. The e-mail subject generation device according to claim 7 or 11, further comprising means for performing. 13. A part-of-speech conversion table for changing a part-of-speech of a word string of the body of the electronic mail to a part-of-speech of the generation pattern, wherein the subject generation unit generates a subject by referring to the part-of-speech conversion table 8. The method according to claim 7, further comprising:
Described email subject generation device. 14. A foreign language pattern pair dictionary having a matching pattern comprising a foreign language word string pattern that can be a subject and a generation pattern comprising a foreign language word string pattern generating the subject, 11. The electronic mail subject generation apparatus according to claim 7, wherein said means includes means for referring to said foreign language pattern pair dictionary when a foreign language word is acquired by said word string decomposition means. 15. A voice conversion unit for converting the subject generated by the subject generation unit into a voice, and a voice output unit outputting the voice converted by the voice conversion unit as it is or to a voice terminal via a network. The electronic mail subject generation device according to claim 7, further comprising: 16. An e-mail acquisition process for inputting a text of an e-mail, a word string decomposition process for decomposing the e-mail into a word string, a collation pattern including a word string pattern that can be a subject, and generating the subject A matching process of referring to a pattern pair dictionary having a generated pattern consisting of a pattern of a word string to be matched, and comparing the matching pattern with the word string decomposed by the word string decomposing process; and a word matched by the matching process. Based on the column,
A subject generation process for generating a subject by referring to a part-of-speech conversion table for changing the part of speech of the word string of the body of the email to the part of speech of the generation pattern. Storage medium. 17. An e-mail, further comprising: a voice conversion process for voice-converting the subject obtained by the subject generation process; and a voice output process for voice-outputting the voice-converted subject when a voice output request is received. A storage medium that stores a subject generation program.