JP5379416B2 - Language processing apparatus and language processing method - Google Patents

Description

  The present invention relates to a language processing apparatus and a language processing method, and more particularly to a language processing apparatus and a language processing method that support adding a new reading to a character string.

  In the conventional kana-kanji conversion system, search system, text-to-speech system, speech recognition system, etc., if you want to give a reading to a word (new word) that is not registered in the dictionary to be used, input the reading to the user It is common to make it.

  The following various methods have been proposed as a method for registering readings of such new words.

  For example, there is a method of taking out words from text data such as a browsed homepage (HP) or e-mail, and updating the database for kana-kanji conversion by giving the user reading (see Patent Document 1).

  In addition, there is a method in which two types of information, text information and voice information, are input to save the user from inputting readings. Specifically, while extracting words that appear to be unregistered words from text information, the voice broadcast on the radio etc. is recognized and converted to text information, looking for places with the same sentence as text information, Reading information of unregistered words is taken out (see Patent Document 2).

  Further, there is a method of searching for predetermined text data or the Internet using a search key as a new word for the user, and creating a material explaining the new word using the search result (Patent Document 3). reference).

  In addition, as a method of displaying the search result, there is a method of displaying a location that matches the search key. For example, a KWIC (KeyWord In Context) search that places the matched location in a specific location and displays the surrounding context. It has been known. KWIC search is also used in general search engines on the Internet, and an HP text excerpt is displayed as a search result, and a portion corresponding to a search key is highlighted in bold. In addition, there is a method in which a query term is searched as a search key and a sentence around a matched portion is displayed (see Patent Document 4).

  In addition, as a method of automatic registration in a dictionary, there is a method of using search results on the Internet as statistical information of words to be registered in a dictionary for language analysis (see Patent Document 5).

By the method as described above, the user can register and use the reading of the new word.
Japanese Patent Laid-Open No. 2002-229999 JP 2005-234219 A Japanese Patent Laid-Open No. 2004-157897 Patent No.03303912 Patent Japanese Patent Laid-Open No. 2002-297587

  The method of registering new readings as needed by the user as shown in the above conventional example can be sufficiently handled if it is performed at the time of document creation by the user, such as a kana-kanji conversion system.

  However, when dealing with news, electronic program guide information, etc. that occur on-time, new words that are not registered in the dictionary often appear and the reading is often difficult to understand. In such a case, it was difficult for the user to input the new word reading accurately and immediately.

  For example, in the method described in Patent Document 1 described above, if the user does not know the reading, it is of course impossible to give the reading, and there is a possibility of giving the wrong reading. In the method described in Patent Document 2, reading cannot be given if there is no audio information for supplementing character information.

  Therefore, it is conceivable that the system automatically creates a reading for an unknown new word, but in this case, there is a problem that the wrong reading is registered and used as it is without the user noticing a reading error. . In order to prevent such erroneous registration, and in consideration of the case where reading different from normal reading is applied, it is desirable to use confirmation by the user together with automatically generated reading.

  Therefore, it is conceivable to confirm the correct reading by performing, for example, an Internet search for new words. When the method described in Patent Document 3 is applied as this search, a search based only on the notation is performed, so the search results include many items that do not have reading information, and are examined one by one. This is a complicated task.

  Further, when displaying the search result, it is desirable to highlight, for example, a portion where the correct reading of the new word is written so that the user can easily recognize it. However, when only the part corresponding to the search key (in this case, a new word) is highlighted in the search result like a general search engine, the part corresponding to the reading that the user wants to know is highlighted. Not. Moreover, even if the method of displaying the sentence surrounding the question term described in Patent Document 4 is used, the portion that is desired to be read is not highlighted.

  In addition, when registering a new word reading as a dictionary, the method of registering a dictionary using search results as statistical information of a new word described in Patent Document 5 uses new word notation information. Because the registration is performed by extracting the connection relationship before and after that, registration limited to reading is not possible.

  In this way, in order to confirm the reading of a new word and register it in the dictionary, when a search is performed using only the new word notation as a key, a lot of information that does not include the reading is also extracted as the search result. Therefore, it is not easy for the user to confirm an accurate reading.

  The present invention has been made to solve the above-described problem, and an object of the present invention is to provide a language processing apparatus and a language processing method capable of efficiently determining an appropriate reading for a character string of a new word. .

  As a means for achieving the above object, a language processing apparatus of the present invention comprises the following arrangement.

That is, an acquisition unit that acquires a character string to be processed, a partial reading generation unit that generates a partial reading character string that indicates a part of the entire reading of the character string, and the character string and the partial reading character string are searched A search means for searching a database including text data as a key, and a complementary reading character string indicating a reading of a portion not corresponding to the partial reading character string in the whole reading of the character string based on a search result by the searching means complement and characterized by having a, a creation means to read the whole to create entire read a string of one or more showing the entire reading said.

  According to the present invention configured as described above, it is possible to efficiently determine appropriate reading for a character string of a new word.

  Hereinafter, the present invention will be described in detail based on preferred embodiments with reference to the accompanying drawings. The configurations shown in the following embodiments are merely examples, and the present invention is not limited to the illustrated configurations.

<First Embodiment>
System Configuration In the information processing system of the present embodiment, the user is supported to give appropriate readings to character strings that are not registered in the dictionary.

  FIG. 1 is a block diagram showing the configuration of the information processing system of this embodiment. In the figure, reference numeral 101 denotes a notation input unit that acquires a character string (hereinafter referred to as a notation character string) that cannot be read. Reference numeral 102 denotes a morpheme analysis dictionary for morphological analysis of a written character string. Reference numeral 103 denotes a morpheme analysis unit that acquires information such as readings and parts of speech of morphemes that form a written character string. Reference numeral 104 denotes a partial reading creation rule for creating a partial reading character string indicating a part of the entire reading of the written character string using the morphological analysis result. A partial reading creation unit 105 creates a partial reading character string by eliminating partial readings from the morphological analysis result using the partial reading creation rule 104. Reference numeral 106 denotes a reading completion rule for detecting a complementary reading character string indicating a reading of an uncertain part not corresponding to the partial reading character string in the entire reading of the written character string from the search result. Reference numeral 107 denotes a search target text in which a written character string and partial reading are searched. A search unit 108 searches the search target text 107 using the written character string and the partial reading character string as search keys. Reference numeral 109 denotes a whole reading detection unit that creates one or more whole reading character strings indicating the whole reading of the written character string by extracting a complementary reading character string from the search result. Reference numeral 110 denotes a result display unit that displays a search result in order to show the user the reading of the written character string. An operation input unit 111 determines an appropriate reading for the written character string based on a user instruction from one or more displayed search results. Reference numeral 112 denotes a dictionary registration unit for registering the final reading for the written character string in the morphological analysis dictionary 102. Note that the morphological analysis dictionary 102, the partial reading creation rule 104, the reading completion rule 106, and the search target text 107 each take the form of a database.

Reading Reading Process FIG. 2 is a flowchart showing a reading search process for a new word in the present embodiment.

  First, in step S201, the notation input unit 101 inputs a notation character string to be read. In step S202, the morpheme analysis unit 103 performs morpheme analysis on the input written character string using the morpheme analysis dictionary 102, and obtains its reading and part of speech.

  Next, in step S203, the partial reading creation unit 105 creates a partial reading character string by excluding a part that is considered indeterminate from the morphological analysis result. This creation process is performed with reference to the partial reading creation rule 104.

  Here, FIG. 3 shows an example of an indeterminate portion exclusion rule in the partial reading creation rule 104. In the present embodiment, classification based on the type of written character string is adopted as an exclusion rule for indeterminate portions. This type is given in advance, for example, by analyzing the format of a text including a notation character string. And the reason for excluding the uncertain part is defined according to the notation type. Reasons for exclusion are: “A. There are words with parts of speech other than proper names of surnames and surnames”, “B. There are multiple readings”, “C. Including unknown words in Kanji” “D. Unknown in alphabets” And "E. Including symbols and numbers". In the example of FIG. 3, there are five types of exclusion reasons A to E for the “person name” type, and four types of exclusion reasons B to E for the “title” type.

  For example, the reason for exclusion A applied to the notation type “person name” is that if “a word with a different part of speech is mixed in addition to the composition of the surname and first name”, an analysis error occurs at the position where the different part of speech is mixed. This indicates that the error is excluded. Therefore, according to this exclusion reason A, either the last name or the last name that can be confirmed is adopted as the partial reading. Specifically, if there is a notation character string “Satomi Sugita”, it is assumed that an analysis result “Sugita (sugi no proper surname) Sa (Suke noun) Tomoe (Chi proper name)” is obtained. In this case, since the noun “sa (noun)” is mixed in the personal name, it is determined that the reading of the name part is inappropriate, and “too much” is adopted as the partial reading.

  Other exclusion reasons B to E applied to the notation type “person name” or “title” are also shown in FIG. 3 and will be briefly described here.

  The reason for exclusion B indicates that, in the case of “a plurality of readings”, a part having a possibility of the plurality of readings is excluded, and the remaining definable readings are adopted as the partial readings. Specifically, if there is a character string “Kokubun Ichiro”, there are two readings in “Kokubun” as the analysis result, so the remaining “Ichirou” is excluded as a partial reading. adopt.

  Further, the exclusion reason C indicates that the unknown word is excluded in the case of “including an unknown word expressed in Kanji”, and the remaining reading is adopted as the partial reading. Specifically, when there is a notation character string “Reino Kawai”, “Reino” is an unknown word as an analysis result. Therefore, “Kawai”, which is the remaining word except for this, is adopted as a partial reading.

  The exclusion reason D indicates that the unknown word is excluded in the case of “including unknown words expressed in alphabets”, and the remaining reading is adopted as the partial reading. Specifically, when there is a character string “New Entrance SuperRookie”, “SuperRookie” is an unknown word as an analysis result.

  The exclusion reason E indicates that the symbol or number is excluded when “includes symbol or number”, and the remaining reading is adopted as the partial reading. Specifically, when there is a notation character string “West Jani ∞”, since “∞” is an unknown word as an analysis result, the remaining “Nishijani” is excluded as a partial reading. .

  In the present embodiment, an example in which partial reading is automatically created based on an exclusion rule has been described. However, partial reading can be determined by an instruction input by a user.

  When the partial reading is created as described above, in step S204, the search unit 108 searches the search target text 107 using the notation character string and the partial reading character string as search keys.

  In step S205, it is determined whether any search result is obtained. If there is no search result, the process advances to step S212 to display that the reference reading cannot be presented because there is no corresponding notation character string and partial reading character string, and the process ends.

  On the other hand, if there is a search result, the process proceeds to step S206, where it is determined whether or not the processing described later has been completed for all the search results, and if there is an unprocessed search result, the process proceeds to step S207.

  In step S207, one of the search results is extracted, and in step S208, the whole reading detection unit 109 identifies the written character string and the reading based on the search result. That is, by extracting the supplementary reading character string that is a portion not corresponding to the partial reading character string in the whole reading for the notation character string from the search result by the written character string and the partial reading character string. , Confirm the entire reading for the notation string. In this determination, a predetermined reading completion rule 106 is applied. Hereinafter, the entire reading for the notation character string determined here is referred to as a whole reading character string. That is, by adding the complementary reading character string extracted here to the partial reading character string, the entire reading character string is determined.

  Here, FIG. 4 shows an example of the reading completion rule 106. That is, the reading completion rule 106 defines a rule for extracting a complementary reading character string based on the positional relationship between the written character string and the partial reading character string included in the search result. Here, in the exclusion rule shown in FIG. 3, the description “Reino Kawai” shown as a specific example of the reason for exclusion C will be described in detail. In this case, “kawai” is generated as a partial reading in step S203, and the search is performed using “no Kawai” and “kawai” as search keys in step S204.

  The detection condition A in FIG. 4 is “when there is a part enclosed in parentheses in the notation, and the inside of the parenthesis is all kana and includes partial reading”, “other than partial reading in kana inside the parenthesis” This indicates that the part is detected as interpolation reading. That is, when there is a kana character string enclosed in parentheses immediately after notation, and the kana string includes partial reading, a portion excluding partial reading becomes an interpolation portion of the reading. In the specific example shown in FIG. 4, “Ayano” in the underlined portion is extracted as a complementary reading character string.

  In the present embodiment, other detection conditions B to E are included as a reading interpolation rule. A specific example thereof is shown in FIG. 4 and will be briefly described here.

  The detection condition B is “other than partial reading in a kana character string including partial reading inside the parenthesis” when “there is a portion enclosed in parentheses in the notation and partial reading is included inside the parenthesis”. This indicates that the location is detected as a complementary reading character string.

  The detection condition C is that when there is a tag notation indicating reading before and after the notation, and there is a kana character string including a partial reading in succession, a part other than “partial reading in the kana character string” is complemented. Indicates that it is detected as a reading character string.

  The detection condition D is that, when “there is a kana character string including a partial reading in a certain section before and after the notation”, the position “before and after partial reading in the kana character string” is detected as a complementary reading character string. Indicates.

  Detection condition E complements the following parts when "There is a part enclosed in parentheses including the notation, followed by a part enclosed in parentheses, and there is a kana character string including partial reading inside the parentheses" Indicates that it is detected as a reading character string. That is, a part of “a reading of characters in front and back of a kana character string inside a parenthesis other than a partial reading is detected” is detected.

  When the entire reading character string is determined as described above, in step S209, the entire reading character string determined based on the current search result does not appear in the previous search results, that is, is created as a candidate. It is determined whether or not it is a new whole reading character string that has not been processed. If it is a new whole reading character string, the process proceeds to step S211, and a new “notation / reading” candidate is created, and the newly created “notation / reading” candidate is currently being processed for the newly created “notation / reading” candidate. Add results. On the other hand, if it is not a new whole reading character string, that is, if it is an already detected whole reading character string, the process proceeds to step S210, and a “notation / reading” candidate already created is currently being processed. Add search results.

  Thereafter, the processing returns to step S206, and it is determined whether or not the processing for all search results has been completed. If completed, the process proceeds to step S212, where a “notation / reading” candidate and a list of search results associated therewith are displayed on the result display unit 110 to be presented to the user, and the process ends.

  Through the above-described reading search process, one or more candidates for new word notation and reading are displayed in a list together with the search results as the basis thereof, so that the user can easily select an appropriate reading. .

Notation Input Processing Here, the notation input unit 101 in this embodiment will be described in detail. FIG. 5 is a block diagram showing a detailed configuration of the notation input unit 101. In FIG. 5, reference numeral 501 denotes a text input unit for inputting text data to be processed. A new word detection unit 502 analyzes input text data and detects a new word. Reference numeral 503 denotes an appearance history database that manages the history of words that have appeared in the text input so far, and has appearance frequency and appearance time information about the words. That is, the notation input unit 101 outputs a word detected as a new word by the new word detection unit 502 to the morpheme analysis unit 103 as a notation character string to be searched for reading.

  Here, the new word detection processing in the new word detection unit 502 of the notation input unit 101 will be described with reference to the flowchart of FIG.

  First, in step S601, text data to be processed is input from the text input unit 501. In step S602, the format of the input text data is analyzed. Specifically, assuming that a description format such as title, performer, genre, and outline is predetermined as in an electronic program guide on television, each item is extracted. For example, performers are cut out for each person name. It should be noted that text data in an atypical format as well as a standard format such as an electronic program guide can be processed. In that case, the unit of a new word is made into a word using a morphological analysis result. Alternatively, the unit of new words may be a phrase using dependency analysis.

  When the extraction of words from the text data input as described above is completed, in step S603, it is determined whether or not a new word determination process described later has been completed for all of the extracted words. If all have been completed, the process ends. If there is an unprocessed word, the process proceeds to step S604.

  In step S604, one of the extracted words is extracted as the current processing target word. Then, by searching the appearance history database 503 using the current processing target word as a key in step S605, it is determined in step S606 whether the current processing target word is a new word. As this determination method, for example, with reference to the appearance history of the current processing target word, it is determined that the current processing target word is a new word as a matter of course when it is not managed yet, that is, there is no appearance history. Also, even if there is an appearance history, if it appears rarely and the last appearance time is old due to the appearance frequency and appearance time information, it is determined that it is necessary to reexamine the reading after the current processing target. That is, it is determined that it is the same as the new word.

  Although an example in which the determination of a new word is performed by referring to the appearance history database 503 is shown here, the present invention is not limited to this example, and may be unregistered with reference to the morphological analysis dictionary 102, for example. It may be determined that the word is a new word.

  If it is a new word, the current processing target word is output to the morpheme analyzing unit 103 in step S607, and is registered in the appearance history database 503 in step S608. On the other hand, if it is not a new word, the appearance frequency and appearance time history are updated in the appearance history database 503 in step S608.

  If the new word determination in step S606 depends only on the presence / absence of the appearance history, it is not necessary to store the appearance frequency and the appearance time. That is, in this case, if it is determined in step S606 that the word is not a new word, the appearance frequency and appearance time history in step S608 may not be updated.

  The above-described processing of steps S604 to S608 is performed on all words extracted from the text data, so that all the new words included in the text data are read and searched in the first embodiment described above. And candidate readings are presented.

  As described above, in the present embodiment, by detecting a new word based on the history of words appearing in the text, it is possible to prevent the reading that needs to be updated from being used as it is old.

Result Display / Registration Processing A display method in the result display unit 110 in this embodiment will be described in detail below.

  In the result display unit 110, when displaying the search result by the notation character string and the partial reading character string, the notation character string given as an input and the complementary reading character string for complementing the partial reading character string are around. It is highlighted with a display form different from other character strings.

  For example, when a notation character string “Reino Kawai” and a partial reading character string “Kawai” are given, the supplementary reading character string corresponding to “Reino” in the associated search result is highlighted. As highlighting in this case, inversion, change of character color, background color, font size, bold display, blinking display, underline, etc. can be considered. In the specific example of the reading interpolation rule shown in FIG. 4, an example of highlighting with an underline is shown.

  Here, FIG. 7 shows another example of highlighted display in the result display unit 110. In this example, when the notation character string “Ichiro Kokubun” and the partial reading character string “ICHIRO” are used as search keys, two types of whole reading character strings “KOKUBUICHIRO” and “KOKUBUICHIRO” are detected. ing. In the present embodiment, for a pair (one candidate) of the same written character string and whole reading character string, one or more corresponding search results are collectively displayed.

  A check box is added to the display of candidates for “notation / reading” on the left side of FIG. The user selects an appropriate whole reading character string for the written character string by selecting any check box from the operation input unit 111, and the instructed candidate is stored in the morphological analysis dictionary 102 by the dictionary registration unit 112. New registration.

In each of the search results displayed in the center of FIG. 7, a notation character string, a partial reading character string, and a complementary reading character string appearing in the text are displayed. In particular, the supplementary reading character string is highlighted by underlining. In the example shown in FIG. 7, the notation character string (Ichiro Kokubun) and the partial reading character string (ICHIRO) used as the search keys are also highlighted in bold, but the supplementary reading character string is clearly Differentiated. Note that the supplementary reading character string or the like may be highlighted in the notation character string and the entire reading character string in the “notation / reading” candidate as in the search result display.
Further, the link destination of each search result is displayed on the right side of FIG. 7, and when the user wants to determine the authenticity of the text in each search result, the contents of the original text can be confirmed by following each link. it can.

  As described above, according to the present embodiment, one or more candidates for the new character notation character string and the whole reading character string are displayed together with the corresponding search results. Furthermore, since the supplementary reading character string is highlighted in the display, the user's viewability is improved, and the user can easily select and register an appropriate reading.

<Modification>
In addition, this invention is not limited to embodiment mentioned above, The modification is also the category of this invention. Hereafter, the modification with respect to this embodiment is given.

In the present embodiment, the reading has been described as a character string of hiragana, but the present invention is not limited to this example, and may be, for example, katakana or romaji.
In the present embodiment, the position information on the notation character string of the partial reading character string is not particularly considered in the reading completion rule. Therefore, matching may be performed using this position information, and the presentation order may be changed or not employed as a search result when there are extra kana characters before and after.

  Moreover, in the reading completion process, an example is shown in which the complementary reading character string extracted from the search result is used as it is. However, it is also effective to examine the accuracy of reading by referring to how to read kanji. . That is, the presentation order may be changed based on the certainty or may not be adopted as a search result.

  Moreover, although the example which employ | adopts only one place in the whole reading character string as a partial reading character string was shown, you may employ | adopt a part with certain reading and use these as a search key. In this case, reading complementation is performed for a plurality of locations.

  In addition, the description character string to be read is described as having been given in advance, but the type (person name, place name, etc.) obtained by morphological analysis using a proper noun dictionary without giving it. May be used.

  In addition, in the partial reading creation rule, a notation type such as “person name” or “title” is used, but an attribute obtained by analyzing the format of text data may be added as a search key. For example, for data having format information such as title, performer, genre, and outline, such as an electronic program guide, first, the occupation of the person's name is obtained from tags such as “appearance /”, “director /”, and “screenplay /”. In order to further narrow down the search results, occupations such as “talent”, “director”, and “screenwriter” may be added as search keys.

  In the present embodiment, the database of the search target text 107 is a search target, but it goes without saying that the search may be performed via a network such as the Internet. In this way, when performing a search via the Internet, a partially read Roman character string may be searched for the description of the URL. In this case, "/", ".", "@", Etc. may be used instead of parentheses in the reading completion rule. For example, it is assumed that a URL “www.xxx.jp/yyy/xxx/youseiouhyou.htm” is detected by creating and searching a reading “ouhyou” from a character string “Masamasa dynasty”. In this case, “youseiouhyou” is cut out using “/” or “.” In the URL as a delimiter of the reading Roman character string, the partial reading character string is excluded, and the reading “yousei” is replaced with the supplementary reading character. Can be a column.

  In addition, an example was shown in which search results are added and displayed for the same “notation / reading” pair (candidate), but multiple search results can be displayed without merging the results for the same “notation / reading” candidate. Each may be displayed with “notation / reading”.

  Further, in the case where there is no search result in step S205 in FIG. 2, an example is shown in which the search result is not found and the process ends. However, in such a case, the search may be repeated again using a partial reading character string that is obtained by adding another reading to the appropriate partial reading or using the inappropriate reading portion as a key. . In this way, when creating another partial reading character string and repeating the search, first create a combination of partial reading character strings of parts that seem to be inappropriate first, and search for all possibilities. Alternatively, the search results may be presented in an order that seems to be certain.

  Note that the morphological analysis dictionary 102 updated according to the present embodiment may be used for reading processing in the speech synthesis system or may be used for creating speech recognition grammars in the speech recognition system.

<Other embodiments>
The present invention can take the form of, for example, a system, apparatus, method, program, or storage medium (recording medium). Specifically, the present invention may be applied to a system composed of a plurality of devices (for example, a host computer, an interface device, a scanner, a web application, etc.), or may be applied to an apparatus composed of a single device. .

  The present invention also provides a software program that implements the functions of the above-described embodiments directly or remotely to a system or apparatus, and the system or apparatus computer reads out and executes the supplied program code. Achieved. The program in this case is a computer-readable program corresponding to the flowchart shown in the drawing in the embodiment.

  Accordingly, since the functions of the present invention are implemented by computer, the program code installed in the computer also implements the present invention. In other words, the present invention includes a computer program itself for realizing the functional processing of the present invention.

  In that case, as long as it has the function of a program, it may be in the form of object code, a program executed by an interpreter, script data supplied to the OS, or the like.

  Recording media for supplying the program include the following media. For example, floppy disk, hard disk, optical disk, magneto-optical disk, MO, CD-ROM, CD-R, CD-RW, magnetic tape, nonvolatile memory card, ROM, DVD (DVD-ROM, DVD- R).

  As a program supply method, the following method is also possible. That is, the browser of the client computer is connected to a homepage on the Internet, and the computer program itself (or a compressed file including an automatic installation function) of the present invention is downloaded to a recording medium such as a hard disk. It can also be realized by dividing the program code constituting the program of the present invention into a plurality of files and downloading each file from a different homepage. That is, a WWW server that allows a plurality of users to download a program file for realizing the functional processing of the present invention on a computer is also included in the present invention.

  In addition, the program of the present invention is encrypted, stored in a storage medium such as a CD-ROM, distributed to users, and key information for decryption is downloaded from a homepage via the Internet to users who have cleared predetermined conditions. It is also possible to make it. That is, the user can execute the encrypted program by using the key information and install it on the computer.

  Further, the functions of the above-described embodiments are realized by the computer executing the read program. Furthermore, based on the instructions of the program, an OS or the like running on the computer performs part or all of the actual processing, and the functions of the above-described embodiments can also be realized by the processing.

  Further, the program read from the recording medium is written in a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, and then executed, so that the program of the above-described embodiment can be obtained. Function is realized. That is, based on the instructions of the program, the CPU provided in the function expansion board or function expansion unit can perform part or all of the actual processing.

It is a block diagram which shows the structure of the information processing system in one Embodiment which concerns on this invention. It is a flowchart which shows the reading search process in this embodiment. It is a figure which shows an example of the exclusion rule at the time of the partial reading character string preparation in this embodiment. It is a figure which shows an example of the complement rule at the time of the whole reading character string preparation in this embodiment. It is a figure which shows the structure of the description input part in this embodiment. It is a flowchart which shows the new word detection process in this embodiment. It is a figure which shows the example of a display of the search result in this embodiment.

Claims (16)

An acquisition means for acquiring a character string to be processed;
Partial reading creation means for creating a partial reading character string indicating a part of the entire reading for the character string;
Search means for searching a database including text data using the character string and the partial reading character string as a search key;
Based on the search result by the search means, a complementary reading character string indicating a reading of a portion not corresponding to the partial reading character string in the entire reading of the character string is complemented, and an entire reading character string indicating the entire reading is obtained. A whole reading creation means for creating one or more ;
Language processing apparatus characterized by having a.   further,
  Output means for outputting one or more whole reading character strings created by the whole reading creation means as reading candidates;
  Selecting means for selecting a reading for the character string from one or more whole reading character strings output by the output means;
The language processing apparatus according to claim 1, further comprising: The language processing apparatus according to claim 2 , wherein the output unit outputs a search result by the search unit in association with each of the one or more whole reading character strings. The output means displays the whole reading character string and the search result in association with each other,
The language processing apparatus according to claim 3 , wherein the selection unit selects a reading for the character string from one or more of the whole reading character strings based on a user instruction. The language processing apparatus according to claim 4 , wherein the output unit highlights the complementary reading character string included in the search result. The output means is different from the character string and the partial reading character string included in the search result, the complementary reading character string included in the search result, and another character string included in the search result. 6. The language processing apparatus according to claim 5 , wherein the language processing apparatus is displayed in a form. Furthermore, it has morpheme analysis means for morphological analysis of the character string,
The partial reading creation means, based on the morphological analysis result of the morphological analysis means, language processing apparatus according to any one of claims 1 to 6, characterized in that to create the partial read character string. Furthermore, morpheme analysis means for morphological analysis of the character string,
Readings selected by the selecting means, have a registration means for registering the morphological analysis dictionary used in the morphological analysis means,
The language processing apparatus according to claim 2, wherein the partial reading creation unit creates the partial reading character string based on a morpheme analysis result by the morpheme analysis unit. The whole reading creation means extracts one or more whole reading character strings by extracting the complementary reading character strings from each of one or more search results by the searching means and complementing the partial reading character strings. language processing apparatus according to any one of claims 1 to 8, wherein the creating. The language according to claim 9 , wherein the whole reading creation unit extracts the complementary reading character string based on a positional relationship between the character string and the partial reading character string included in the search result. Processing equipment. Furthermore, it has a history management means for managing the appearance history of words in text data,
The language according to any one of claims 1 to 10 , wherein, when a word that has not yet been managed by the history management unit appears, the acquisition unit acquires the word as the character string. Processing equipment. Said retrieval means, language processing apparatus according to any one of claims 1 to 11, characterized in that a search through the Internet. A language processing method performed by a language processing device,
An acquisition step in which the acquisition unit of the language processing apparatus acquires a character string to be processed;
A partial reading creation step in which the partial reading creation means of the language processing device creates a partial reading character string indicating a part of the entire reading for the character string;
A search step in which the search means of the language processing device searches a database including text data using the character string and the partial reading character string as search keys;
The whole reading creation means of the language processing device complements a supplementary reading character string indicating a reading of a portion not corresponding to the partial reading character string in the whole reading for the character string, based on a search result by the searching step. A whole reading creation step for creating one or more whole reading character strings indicating the whole reading ;
Language processing method characterized by having a.   further,
  An output step in which the output means of the language processing device outputs one or more whole reading character strings created in the whole reading creation step as reading candidates;
  A selection step in which the selection unit of the language processing device selects a reading for the character string from one or more of the whole reading character strings output in the output step;
The language processing method according to claim 13, further comprising: A program for causing a computer to function as the language processing apparatus according to any one of claims 1 to 12 . A computer-readable storage medium storing the program according to claim 15 .

Images