Device and program for detecting spelling variations

Abstract

To provide a notational variation detection apparatus and program for detecting notational variations in a document and presenting correction candidates using a plurality of documents.SOLUTION: A document verification server 1 comprises: a notational variation word identifying unit 13 for identifying, as a notational variation word, a word that matches a candidate word in a notational variation dictionary 23 among words included in a verification target document; a candidate word extraction unit 14 for extracting a plurality of candidate words included in a set of a plurality of candidate words for a word related to the notational variation of the notational variation dictionary 23 that includes the identified notational variation word; an accumulated document search unit 16 for searching for a word that matches any of the plurality of candidate words extracted by the candidate word extraction unit 14 from a plurality of registered documents stored in a document storage unit 26 and searching for a character string that includes a word whose context is the same as the notational variation word identified from the searched word; and a list generation unit 17 for aggregating the words and character strings searched by the accumulated document search unit 16 for each candidate word, and generating a candidate word list in which the aggregated numbers are associated with the candidate words.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a spelling variation detection device and a program. [Background technology]

[0002] Traditionally, companies have set up customer inquiry pages on their websites to answer customer inquiries. These inquiry-related tasks are sometimes outsourced to specialized companies. These specialized companies, commissioned by companies, create answers to customer inquiries by searching manuals and frequently asked questions (FAQs). However, if there are variations in the way answers are written, they can miss certain parts of the answer or the answers created can be difficult to read. In view of this situation, for example, a variation word determination device that can accurately extract expressions that result in spelling variations in a document has been disclosed (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6413659 Summary of the Invention [Problem to be solved by the invention]

[0004] The technique described in Patent Document 1 determines whether a word in a document is a variation word that represents a variation in orthography. Regarding variations in spelling, it is important not only to standardize spelling variations within a document, but also to standardize spelling variations with other documents, such as other responses to customer inquiries about the same company, to prevent oversight during searches, etc. In particular, when multiple people are creating responses to customer inquiries, it is desirable to use the same spelling regardless of who creates them. On the other hand, some names used in product names, etc., need to be written in a fixed way, and it would be problematic to unify all names that are judged to have spelling variations.

[0005] SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide a spelling variation detection device and program that detects spelling variations in a document and presents correction candidates using multiple documents. [Means for solving the problem]

[0006] The present invention solves the above problems by the following means. The first invention is a spelling variation detection device for detecting spelling variations in a document, comprising: a spelling variation dictionary storing a plurality of pairs of candidate words for a word associated with one spelling variation; a spelling variation word specification means for specifying, as a spelling variation word, a word contained in a document to be verified that matches a candidate word in the spelling variation dictionary; a candidate word extraction means for extracting a plurality of candidate words contained in the pair of candidate words in the spelling variation dictionary that includes the spelling variation word specified by the spelling variation word specification means; and a document database storing a plurality of registered documents stored in the document database. The spelling variation detection device comprises: a document search means for searching for a word that matches any of the plurality of candidate words extracted by a column; a context search means for searching, from the words searched for by the document search means, for a string that includes a word that has the same context as the spelling variation word identified by the spelling variation word identification means; a usage counting means for counting the words searched for by the document search means for each candidate word and counting the strings searched for by the context search means for each candidate word; and a candidate word output means for outputting the numbers counted by the usage counting means in association with the candidate words. The second invention is a spelling variation detection device according to the first invention, further comprising a document output means for outputting the document to be verified and highlighting the spelling variation words identified by the spelling variation word identification means. The third invention is a spelling variation detection device according to the second invention, wherein the document output means outputs the spelling variation words with different emphasis modes depending on the number of each candidate word counted by the usage counting means or the ratio of the number of each candidate word to the total number. A fourth invention is a spelling variation detection device according to the second or third invention, wherein the candidate word output means outputs an operation member related to the modification of the candidate word, and includes a document update means, when an operation related to the operation member is received, that updates the spelling variation word in the document to be verified output by the document output means to the candidate word corresponding to the operation member. A fifth invention is a spelling variation detection device according to any one of the first to fourth inventions, wherein the usage counting means counts the number of character strings searched by the context search means. The sixth invention is a spelling variation detection device according to any one of the first to fourth inventions, wherein the usage counting means counts the number of registered documents that contain the character string searched by the context search means. A seventh invention is a spelling variation detection device according to any one of the first to sixth inventions, wherein the candidate word output means further outputs the ratio of the number of each candidate word to the total number of searched words. The eighth invention is a spelling variation detection device according to any one of the first to seventh inventions, further comprising a part-of-speech acquisition means for acquiring the parts of speech of the spelling variation word identified by the spelling variation word identification means and of the words before and after the spelling variation word in the document to be verified, and the context search means searches for the string containing the word searched by the document search means, whose parts of speech of the word and the words before and after the word match each of the parts of speech acquired by the part-of-speech acquisition means. The ninth invention is a spelling variation detection device according to any one of the second to fourth inventions, comprising a unified rule memory unit that stores a unified word for a word, a unified rule confirmation means that confirms whether the spelling variation word identified by the spelling variation word identification means matches the word in the unified rule memory unit, and a modified word output means that, when the unified rule confirmation means confirms a match, outputs the unified word corresponding to the matched word instead of the output by the candidate word output means. The tenth invention is a spelling variation detection device according to the ninth invention, wherein the document output means highlights and outputs the spelling variation words that the unified rule confirmation means has confirmed to match in a manner different from the output by the candidate word output means. The eleventh invention is a spelling variation detection device according to any one of the first to tenth inventions, further comprising a document registration means for registering the verification target document after addressing the spelling variation words in the document database. A twelfth invention is a spelling variation detection device according to any one of the first to eleventh inventions, comprising a morphological analysis means for performing morphological analysis on the document to be verified, and a word extraction means for extracting the words from the morphological analysis results by the morphological analysis means. A thirteenth aspect of the present invention is a program for causing a computer to function as any one of the spelling variation detection devices according to the first to twelfth aspects of the present invention. [Effects of the Invention]

[0007] According to the present invention, it is possible to provide a spelling variation detection device and program that detects spelling variations in a document and presents correction candidates using a plurality of documents. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is an overall configuration diagram of a document verification system according to an embodiment of the present invention and a functional block diagram of a document verification server. [Figure 2] FIG. 4 is a diagram illustrating an example of a storage unit of the document verification server according to the embodiment. [Figure 3] 10 is a flowchart illustrating a document verification process of the document verification server according to the present embodiment. [Figure 4] FIG. 10 is a diagram showing an example of a verification screen output to a user terminal according to the embodiment. [Figure 5] 10 is a flowchart illustrating a document analysis process of the document verification server according to the present embodiment. [Figure 6]10A and 10B are diagrams illustrating a specific example of a document analysis process of the document verification server according to the embodiment. [Figure 7] 10A and 10B are diagrams illustrating a specific example of a document analysis process of the document verification server according to the embodiment. [Figure 8] 10A and 10B are diagrams illustrating a specific example of a document analysis process of the document verification server according to the embodiment. [Figure 9] FIG. 10 is a diagram showing an example of a verification screen output to a user terminal according to the embodiment. [Figure 10] FIG. 10 is a diagram showing an example of another verification screen output to the user terminal according to the embodiment. [Figure 11] 10 is a flowchart showing post-output processing of the document verification server according to the present embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, this is merely an example, and the technical scope of the present invention is not limited to this example. (Embodiment) <Overall Configuration of Document Verification System 100> FIG. 1 is a diagram showing the overall configuration of a document verification system 100 according to this embodiment and a functional block diagram of a document verification server 1. As shown in FIG. FIG. 2 is a diagram showing an example of the storage unit 20 of the document verification server 1 according to this embodiment.

[0010] 1, the document verification system 100 includes a document verification server 1 (a spelling variation detection device) and a user terminal 5. The document verification server 1 and the user terminal 5 are connected via a communication network N. In the document verification system 100, for example, the document verification server 1 receives and analyzes a response sentence (verification target document) input from the user terminal 5 in response to a question from a customer. Then, the document verification server 1 outputs to the user terminal 5 the response sentence in which the spelling variations are highlighted, as an analysis result of the spelling variations, and a candidate word list indicating the usage situation related to the application of the spelling variations in other documents.

[0011] In this way, the document verification system 100 is a system that outputs the analysis results regarding spelling variations in the answer sentence to the answer sentence creator (hereinafter referred to as the user), and provides the user with information to determine whether or not to correct the spelling variations. In the following description, the document to be verified will be described using, as an example, a reply written by a user in response to a question from a customer, but the document verification system 100 can be used for various documents.

[0012] <Document Verification Server 1> The document verification server 1 is a server that verifies the answer sentence and presents the verification results. More specifically, the document verification server 1 detects spelling variations in the answer sentence by analyzing the answer sentence. Furthermore, the document verification server 1 outputs candidate words for the detected spelling variations based on the context of the accumulated registered documents. The document verification server 1 includes a control unit 10, a storage unit 20, and a communication interface unit 29. The control unit 10 is a CPU (Central Processing Unit) that controls the entire document verification server 1. The control unit 10 appropriately reads and executes an OS (Operating System) and various application programs stored in the storage unit 20, thereby cooperating with the above-mentioned hardware and executing various functions.

[0013] Before describing each function of the control unit 10, the storage unit 20 will be described. The storage unit 20 is a storage device such as a hard disk or semiconductor memory device for storing programs, data, etc. required for the operation of the document verification server 1. The storage unit 20 includes a program storage unit 21, a spelling variation dictionary 23, a unified rule storage unit 24, and a document storage unit 26 (document database).

[0014] The program storage unit 21 is a storage area for storing various programs executed by the document verification server 1. The program storage unit 21 stores a verification program 21a (program). The verification program 21a is, for example, a program for executing various functions of the control unit 10, which will be described later. The programs for executing the various functions of the control unit 10 do not have to be realized by one verification program 21a as described above. For example, a program may be provided for each function of the control unit 10, or for each of multiple functions.

[0015] The spelling variation dictionary 23 is a storage area for storing candidate words for spelling variations. FIG. 2(A) shows an example of the spelling variation dictionary 23. The spelling variation dictionary 23 shown in FIG. 2(A) stores a plurality of candidate words (candidate word 1, candidate word 2, . . . ) in association with each other, using a spelling variation ID (IDentification) as a key. The spelling variation ID is identification information that identifies a set of words that represent the same word. The multiple candidate words are two or more, and the number varies depending on the word. For example, the spelling variation ID "1" includes multiple candidate words such as "uaiawase," "uaiwaise," and "uaiawase," as well as "uaiawase," for example.

[0016] The unified rule storage unit 24 stores the unified character string in the document in association with the detected character string. FIG. 2B shows an example of the unified rule storage unit 24. The unified rule storage unit 24 shown in FIG. 2(B) stores the type, the detected character string, the part of speech, the corrected character string, and the display content in association with each other. The type is either correction or confirmation. Correction indicates that correction is necessary. Confirmation indicates that confirmation is required as to whether correction is necessary. The detection string is a string to be detected from the document. The part of speech is the part of speech of the character string to be detected. Note that the part of speech is an optional item. The correction string is a string that replaces the detection string and is a unified word. The display content is the content that is output to the screen as the verification result. For example, if the word "refund" is found in the document and its part of speech is a verb, "refund" is output as the corrected character string, and the display content is also output.

[0017] It should be noted that, for example, a different unified rule storage unit 24 may be prepared for each company. Furthermore, the unified rule storage unit 24 may include, although not limited to spelling variations, for example, words that should be unified such as service names, and prohibited words such as in-house abbreviations. 1 is a storage area for storing a plurality of documents. The document storage unit 26 stores, for example, dealt with answer sentences (registered documents) that have been dealt with in response to spelling variations. Note that, for example, a question sentence may be associated with the dealt with answer sentences and stored in the document storage unit 26.

[0018] Next, the control unit 10 will be described. The control unit 10 includes a document receiving unit 11, a morphological analysis processing unit 12 (morphological analysis means, word extraction means), a spelling variation word identification unit 13 (spelling variation word identification means), a candidate word extraction unit 14 (candidate word extraction means), a unified rule confirmation unit 15 (unified rule confirmation means), a stored document search unit 16 (part of speech acquisition means, document search means, context search means), a list generation unit 17 (usage aggregation means, candidate word output means, modified word output means), a result output unit 18 (candidate word output means, modified word output means, document output means), and a document update registration unit 19 (document update means, document registration means).

[0019] The document receiving unit 11 receives, from the user terminal 5, for example, an input of a reply sentence that is a reply to a question sentence from a customer. The morphological analysis processing unit 12 performs morphological analysis on the answer sentence received by the document receiving unit 11, thereby dividing the answer sentence into the smallest meaningful elements. Then, the morphological analysis processing unit 12 extracts words from each element of the divided answer sentence. Here, the morphological analysis processing unit 12 may extract words whose parts of speech are nouns, verbs, or adjectives, for example. These parts of speech are based on the empirical rule that they are words that are subject to spelling variations.

[0020] The spelling variation word identifying unit 13 identifies, from among the words extracted by the morphological analysis processing unit 12, words that match candidate words in the spelling variation dictionary 23 as spelling variation words. The candidate word extracting unit 14 extracts from the spelling variation dictionary 23 a plurality of candidate words contained in a set of the spelling variation dictionary 23 that includes the spelling variation word identified by the spelling variation identifying unit 13 . The unified rule checking unit 15 checks whether the spelling variation word specified by the spelling variation specifying unit 13 matches the detected character string (word) in the unified rule storage unit 24. The unified rule checking unit 15 also checks whether the spelling variation word specified by the spelling variation specifying unit 13 and its part of speech match the detected character string (word) in the unified rule storage unit 24 and its part of speech.

[0021] The stored document search unit 16 searches the multiple addressed answer sentences stored in the document storage unit 26 for a word that matches any of the multiple candidate words extracted by the candidate word extraction unit 14. The stored document search unit 16 also searches the searched words for character strings containing words that have the same context as the spelling variation word identified by the spelling variation word identification unit 13. Here, "same context" means that the semantic relationship between the spelling variation word and the words before and after it is the same. More specifically, the stored document search unit 16 acquires the parts of speech of the spelling variation word identified by the spelling variation word identification unit 13 and the words before and after the spelling variation word in the answer sentence. The stored document search unit 16 then searches the multiple addressed answer sentences stored in the document storage unit 26 for character strings containing candidate words whose parts of speech match each of the acquired parts of speech of the candidate word and the words before and after the candidate word.

[0022] The list creation unit 17 counts the words and character strings searched by the stored document search unit 16 for each candidate word, and creates a candidate word list in which the counted numbers are associated with the candidate words. At this time, the list creation unit 17 counts the numbers of words and character strings searched by the stored document search unit 16 for each candidate word. The list generating unit 17 may tally up the number of documents of dealt with answer sentences containing the words and character strings searched by the stored document searching unit 16 for each candidate word. The list generating unit 17 may further calculate the ratio of the number of each candidate to the total number searched. Here, the number refers to the number of words and / or character strings that are candidate words. Furthermore, when the unified rule checking unit 15 checks for a match, the list generating unit 17 generates a correction word list including a correction character string (unified word) corresponding to the matched detected character string.

[0023] The result output unit 18 highlights the spelling variations identified by the spelling variation identification unit 13 for the response sentence received by the document receiving unit 11 and outputs the result to the user terminal 5. In this case, if the unified rule checking unit 15 confirms a match, the result output unit 18 highlights the spelling variations that have been confirmed to match in a manner different from that of other spelling variations. Examples of highlighting methods include underlining, highlighting by shading, making the text bold, or making the font size larger than the others. Furthermore, the result output unit 18 outputs the candidate word list and the modified word list generated by the list generation unit 17 to the user terminal 5.

[0024] When the document update registration unit 19 receives an operation related to correction from the candidate word list or the correction word list, it updates the spelling variation word in the response sentence to a word corresponding to the operation. Furthermore, when the document update registration unit 19 receives an operation for registering an answer sentence, it registers the answer sentence being displayed in the document storage unit 26 in association with the question sentence. The details of each of these functions will be described later.

[0025] The communication interface unit 29 is an interface for communicating with the user terminal 5 via the communication network N. Here, a computer refers to an information processing device equipped with a control unit, a storage device, etc., and the document verification server 1 is an information processing device equipped with a control unit 10, a storage unit 20, etc., and is included in the concept of a computer. There is no limit to the number of pieces of hardware that constitute the document verification server 1. It may be configured with one or more pieces as needed. The hardware of the document verification server 1 may be configured to include various servers such as a web server, a DB (database) server, and an application server as needed, and may be configured with one server or with separate servers. The document verification server 1 may also be, for example, a cloud.

[0026] <User terminal 5> 1 is a terminal used by a user who creates a response to a question from a customer, for example. The user terminal 5 can be configured as, for example, a personal computer (PC), a tablet terminal, etc. Although not shown, the user terminal 5 includes a control unit, a storage unit, a display unit, an input unit, a communication interface unit, etc. The communication network N is a network between the document verification server 1 and the user terminal 5, and is, for example, a communication network such as the Internet. The communication network N may be a LAN (Local Area Network) or the like. The communication network N may be wired or wireless.

[0027] <Processing Description> Next, the process performed by the document verification server 1 will be described. FIG. 3 is a flowchart showing the document verification process of the document verification server 1 according to this embodiment. FIG. 4 is a diagram showing an example of a verification screen 50 output to the user terminal 5 according to this embodiment. FIG. 5 is a flowchart showing the document analysis process of the document verification server 1 according to this embodiment. 6 to 8 are diagrams showing a specific example of the document analysis process of the document verification server 1 according to this embodiment. FIG. 9 is a diagram showing an example of a verification screen 80 output to the user terminal 5 according to this embodiment. FIG. 10 is a diagram showing an example of another verification screen 80-2 output to the user terminal 5 according to this embodiment. FIG. 11 is a flowchart showing post-output processing of the document verification server 1 according to this embodiment.

[0028] First, the user terminal 5 connects to the document verification server 1 and requests a verification screen, and in step S (hereinafter, "step S" will be simply referred to as "S") 11 of FIG. 3, the control unit 10 of the document verification server 1 transmits the verification screen to the user terminal 5. At the user terminal 5, the user uses the output verification screen to input a response to the question from the customer.

[0029] FIG. 4 shows a verification screen 50 output to the user terminal 5. The verification screen 50 includes a question sentence 51 , an answer sentence input area 52 , an analysis button 53 , an analyzed answer sentence area 54 , and an analysis result area 55 . The question 51 is a question from a customer and cannot be edited. The answer input area 52 is an area where the user can input an answer to the question 51. The analysis button 53 is an instruction button selected by the user when analyzing the answer sentence input in the answer sentence input area 52. The analyzed answer sentence area 54 and the analysis result area 55 are areas that are output after the answer sentence is analyzed. Therefore, when the verification screen 50 is output to the user terminal 5, nothing is output.

[0030] The user inputs an answer to the question in the answer input area 52 while looking at the question displayed in the question 51. Then, when the user finishes inputting the answer sentence in the answer sentence input area 52 and selects the analysis button 53, the control unit 10 (document receiving unit 11) receives the answer sentence in S12 of FIG. In S13, the control unit 10 performs a document analysis process.

[0031] Here, the document analysis process will be described with reference to FIG. In S21 of FIG. 5, the control unit 10 (morphological analysis processing unit 12) of the document verification server 1 performs morphological analysis processing on the received answer sentence to extract words from the answer sentence. In S22, the control unit 10 (the spelling variation specifying unit 13) compares the words of the extracted response sentence with the candidate words in the spelling variation dictionary 23, and specifies the words that match the candidate words as spelling variations. In S23, the control unit 10 (candidate word extraction unit 14) extracts a plurality of candidate words contained in the set of the spelling variation dictionary 23 that includes the identified spelling variation word.

[0032] In S24, the control unit 10 (unified rule confirmation unit 15) determines whether or not a word identical to the identified spelling variation word exists in the detected character string in the unified rule storage unit 24. If a word identical to the spelling variation word exists (S24: YES), the control unit 10 proceeds to S25. On the other hand, if a word identical to the spelling variation word does not exist (S24: NO), the control unit 10 proceeds to S26. This process is performed on all spelling variations identified from the response sentence. Therefore, if there are multiple spelling variations identified, the determination process is performed on one of them. Then, this process is performed on all spelling variations.

[0033] In S25, the control unit 10 (list generating unit 17) generates a list of corrective words for the spelling variations. After that, the control unit 10 moves the process to S29. On the other hand, in S26, the control unit 10 (stored document search unit 16) acquires the parts of speech of the orthographic variation word and the words (morphemes) before and after it. The control unit 10 can use the parts of speech obtained by morphological analysis in this process.

[0034] In S27, the control unit 10 (stored document search unit 16) searches for a word that matches one of the multiple candidate words extracted from the dealt with answer sentences in the document storage unit 26. Then, the control unit 10 (stored document search unit 16) searches for a string containing a word with an equivalent context based on the parts of speech of the spelling variation word and the words before and after it from the searched word. Here, "equal context" means that the string of the dealt with answer sentences in the document storage unit 26 is one of the multiple candidate words extracted in the processing of S23, the part of speech of the string is the same as that of the spelling variation word, and the parts of speech of the words before and after the string are the same as those of the words before and after the spelling variation word.

[0035] In S28, the control unit 10 (list generation unit 17) counts, for each candidate word, character strings that include words that match any of the searched candidate words and words that have the same context, and generates a candidate word list. In S29, the control unit 10 determines whether all the identified spelling variations have been processed. If all the identified spelling variations have been processed (S29: YES), the control unit 10 proceeds to S14 in Fig. 3. On the other hand, if all the identified spelling variations have not been processed (S29: NO), the control unit 10 proceeds to S24 and performs similar processing on the remaining spelling variations.

[0036] A specific example of the document analysis process will be described with reference to FIGS. The control unit 10 (morphological analysis processing unit 12) of the document verification server 1 performs morphological analysis on, for example, an input sentence 61 shown in Fig. 6(A) to obtain a morphological analysis result 62 shown in Fig. 6(B). The morphological analysis result 62 is obtained by dividing the input sentence 61 into words, and each word is associated with a part of speech. Then, the control unit 10 (morphological analysis processing unit 12) extracts words from the morphological analysis result 62 (S21 in Fig. 5). The control unit 10 extracts, for example, "tomorrow," "meeting," "target," "user," "selection," and "meeting" from the morphological analysis result 62 of FIG. 6(B).

[0037] Next, the control unit 10 (the spelling variation identification unit 13) identifies spelling variations from the extracted words (S22 in FIG. 5). In Fig. 6(B), words 62a to 62c respectively match the candidate words with spelling variation IDs "1" and "3" which are part of the spelling variation dictionary 23 shown in Fig. 6(C). Therefore, the control unit 10 identifies words 62a to 62c as spelling variation words. The control unit 10 (candidate word extraction unit 14) extracts multiple candidate words having a spelling variation ID of "1" in Fig. 6(C) as corresponding to the words 62a and 62c in Fig. 6(B). The control unit 10 also extracts multiple candidate words having a spelling variation ID of "3" in Fig. 6(C) as corresponding to the word 62b in Fig. 6(B) (S23 in Fig. 5).

[0038] The control unit 10 (unified rule checking unit 15) checks whether there is a match in the character strings detected in the unified rule storage unit 24 for the words 62a to 62c shown in Fig. 6(B) (S24 in Fig. 5). In this example, there is no match in the character strings detected in the unified rule storage unit 24 (NO in S24 in Fig. 5), so the control unit 10 (stored document search unit 16) extracts a context object 63 consisting of the spelling variation word "uaiawase" ("meeting") and its part of speech, and the parts of speech of the words (morphemes) before and after the spelling variation word, for the words 62a and 62c shown in Fig. 6(B) (S26 in Fig. 5), as shown in Fig. 7. In addition, the control unit 10 has already extracted a set 23a in the spelling variation dictionary 23 that includes the spelling variation word (S23 in Fig. 5). Therefore, the control unit 10 (stored document search unit 16) refers to the document storage unit 26 and searches the dealt with answer sentences for all words that contain any of the candidate words in the set 23a, and further searches for all character strings that are equal to the context target 63 (S27 in Figure 5).

[0039] 8, the control unit 10 (list generation unit 17) creates a context tally 65 that tallies the number of appearances for the context object 63. The control unit 10 also calculates the ratio for each context and updates the context tally 65 to a ratio tally 66 that includes the ratio. Then, the control unit 10 (list generating unit 17) generates the candidate word list 70 based on the information of the percentage tally 66 (S27 and S28 in FIG. 5). The candidate word list 70 is in the form of a table showing the number and percentage of candidate words contained in all dealt with answer sentences stored in the document storage unit 26, and the number and percentage of candidate words in the same context as the current input sentence 61. The candidate word list 70 also has unify buttons 71a to 71c (operation members) for unifying any of the spelling variations in the input sentence 61 with one of the candidate words. The candidate word list 70 also has correction buttons 72a to 72d (operation members) for correcting the context of the input sentence 61 among the candidate words in the same context as the current input sentence 61.

[0040] In S14 of FIG. 3, the control unit 10 (result output unit 18) performs emphasis adding processing, for example, adding highlighting to the orthographic variation words included in the answer sentence. In S15, the control unit 10 (result output unit 18) outputs to the user terminal 5 the verification screen 80 in which the analysis results have been updated. FIG. 9 shows an example of a verification screen 80 output to the user terminal 5. As shown in FIG. The verification screen 80 is a screen that is updated and output after the analysis button 53 is selected on the verification screen 50 (FIG. 4). The verification screen 80 includes an answer sentence input area 82 , an analyzed answer sentence area 84 , an analysis result area 85 , and a save button 89 .

[0041] The analyzed answer sentence area 84 is an area where an analyzed answer sentence is output in response to the answer sentence entered in the answer sentence input area 82. The analyzed answer sentence reflects the results of the analysis process for the answer sentence entered in the answer sentence input area 82. For example, spelling variations are highlighted so that the document verification results can be seen. In this example, spelling variations such as "announce" and "order" are highlighted in green, for example. Furthermore, parts that require confirmation through other verification processes, such as consecutive punctuation marks, are highlighted in light blue, for example, and detected character strings in the unified rule storage unit 24 are highlighted in yellow because they require correction. Therefore, by referring to the analyzed answer sentence output in the analyzed answer sentence area 84, the verification results, such as the verification results, can be confirmed by the highlight color.

[0042] The analysis result area 85 is an area where the details of the analysis are output. The analysis result area 85 has a confirmation and correction output section 86 and a spelling variation output section 87. The confirmation and correction output unit 86 outputs a list of correction words, which are confirmation and correction items other than spelling variations. In the confirmation and correction output unit 86, the confirmation or correction items are highlighted in the same color as the highlighting applied to the corresponding part of the analyzed answer sentence. The correction items also have an apply button 86a. When the user selects the apply button 86a, the corresponding part of the analyzed answer sentence is updated with the corrected string. The spelling variation output unit 87 outputs a candidate word list. The save button 89 is an instruction button selected by the user when saving the post-analysis answer sentence after making corrections or the like to the answer sentence.

[0043] The verification screen 80 shown in FIG. 9 is an example. For example, as in the verification screen 80-2 shown in FIG. 10, an analysis result area 85-2 may have an analysis result output section 88-2 and a spelling variation confirmation button 90. The analysis result output unit 88-2 outputs the analysis result including spelling variations, and includes a list of corrected words. The spelling variation confirmation button 90 is an instruction button for outputting a candidate word list screen 91 in response to a selection operation by the user. In S16 of FIG. 3, the control unit 10 performs post-output processing.

[0044] Here, the post-output processing will be described with reference to FIG. In S41 of Fig. 11, the control unit 10 determines whether or not a selection operation of the Apply button in the correction word list has been accepted. For example, when the user selects the Apply button 86a (Fig. 9) of the confirmation correction output unit 86, the control unit 10 determines that a selection operation of the Apply button in the correction word list has been accepted. If a selection operation of the Apply button has been accepted (S41: YES), the control unit 10 proceeds to S42. On the other hand, if a selection operation of the Apply button has not been accepted (S41: NO), the control unit 10 proceeds to S43. In S42, the control unit 10 (document update registration unit 19) updates the corresponding portion of the reply sentence to the corrected character string corresponding to the apply button in the corrected word list.

[0045] In S43, the control unit 10 determines whether a selection operation on the candidate word list has been accepted. For example, if any of the Unify buttons 71a to 71c or any of the Modify buttons 72a to 72d has been accepted in the candidate word list 70 of FIG. 8, it is determined that a selection operation on the candidate word list has been accepted. If a selection operation on the candidate word list has been accepted (S43: YES), the control unit 10 proceeds to S44. On the other hand, if a selection operation on the candidate word list has not been accepted (S43: NO), the control unit 10 proceeds to S45. In S44, the control unit 10 (document update and registration unit 19) updates the corresponding portion of the reply sentence to the word corresponding to the selection operation in the candidate word list.

[0046] In S45, the control unit 10 determines whether or not to save the answer sentence. For example, when the user selects the save button 89 (FIG. 9), the control unit 10 determines to save the answer sentence. If the answer sentence is to be saved (S45: YES), the control unit 10 moves the process to S46. On the other hand, if the answer sentence is not to be saved (S45: NO), the control unit 10 moves the process to S41 and continues to make other corrections. In S46, the control unit 10 (document update registration unit 19) stores the updated answer sentence in the document storage unit 26 in association with the question sentence, for example. Thereafter, the control unit 10 ends this process.

[0047] In this way, the document verification server 1 presents candidates for spelling variations as a candidate word list, and the user can correct the answer sentence by giving instructions for correction from the candidate word list, etc. Furthermore, the corrected answer sentence and the answer sentence confirmed by the user are stored in the document storage unit 26, so that the next time a new answer sentence is verified, candidates for spelling variations can be presented that reflect the content of the current correction.

[0048] As described above, according to this embodiment, the document verification server 1 has the following advantages. (1) A spelling variation dictionary 23 is referenced, which stores multiple sets of candidate words for a word associated with one spelling variation, and words contained in the answer sentence, which is the document to be verified, that match a candidate word in the spelling variation dictionary 23 are identified as spelling variation words. Then, multiple candidate words contained in the set in the spelling variation dictionary 23 that includes the identified spelling variation word are extracted, a word that matches any of the extracted candidate words is searched for among multiple addressed answer sentences stored in the document storage unit 26, character strings that include candidate words with the same context as the spelling variation word are searched for, the searched words and character strings are tallied for each candidate word, and a candidate word list is output in which the tallied numbers are associated with the candidate words. Therefore, the candidate words for correcting spelling variations are output as a list of candidate words by aggregating the contexts of the previously addressed answer sentences, allowing the user to refer to the trends in the contexts in which the same spelling variation has been addressed in the past. As a result, support can be provided for deciding how to correct spelling variations in answer sentences.

[0049] (2) The answer sentence is output, and the identified spelling variations are highlighted. Therefore, orthographic variations in the answer sentence are extracted and output with emphasis, so that orthographic variations in the answer sentence can be clearly indicated to the user. (3) In the candidate word list, a button for correcting the candidate word is output, and when an operation related to the button is accepted, the orthographic variation of the answer sentence is updated to the candidate word corresponding to the button. Therefore, the user can conveniently correct the spelling variation to the candidate word selected by the user simply by selecting a button.

[0050] (4) The number of words and character strings found for words that match one of the candidate words for the spelling variation and for character strings that contain candidate words with the same context as the spelling variation are tallied, so the total number output in the candidate word list can be the number of words and character strings contained in the addressed answer sentence. (5) For words that match one of the candidate words for the spelling variation and character strings that include candidate words with the same context as the spelling variation, the total number is calculated based on the number of dealt with answer sentences that include the word and character string. Therefore, the total number output to the candidate word list can be the number of dealt with answer sentences, and can be made to be unaffected by the number of words and character strings included in the dealt with answer sentences. (6) The candidate word list further outputs the ratio of the number of each candidate word to the total number of searched words, so that the user can be provided with information including the usage status of each candidate word in the addressed answer sentences, including the ratio.

[0051] (7) The parts of speech of the identified spelling variation word and the words before and after the spelling variation word in the answer sentence are obtained, and a string containing a candidate word whose parts of speech of the candidate word and the words before and after the candidate word match each of the obtained parts of speech is searched from the multiple addressed answer sentences stored in the document storage unit 26. Therefore, candidate words that have the same part of speech as the orthographic variation word and the parts of speech of the words before and after it can be determined to be character strings with the same context.

[0052] (8) It is confirmed whether the identified spelling variation word matches a word in the unified rule memory unit 24, which stores unified words for words, and if a match is confirmed, instead of outputting a list of candidate words, it outputs a list of modified words including unified words corresponding to the matched word. Furthermore, in the response sentence, the identified spelling variation words are output in a highlighted manner different from the output of the candidate word list. Therefore, for example, product names and the like that are predetermined to be replaced with fixed words are registered in the unified rule storage unit 24, and when a word stored in the unified rule storage unit 24 is included in a reply sentence, a list of modified words is output, making the content of the correction clear. Also, words with spelling variations in the reply sentence are output in a highlighted manner that differs from the output of the candidate word list, making it easy to distinguish between the analysis results for the words.

[0053] (9) The response sentence after dealing with the spelling variation is registered in the document storage unit 26, so that the next time a spelling variation is checked, the response sentence can be included in the count and counted. (10) Morphological analysis is performed on the answer sentence, and words are extracted from the results of the morphological analysis, so words can be extracted from the answer sentence by processing. The parts of speech of the words can also be obtained.

[0054] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments. Furthermore, the effects described in the embodiments are merely a list of the most preferable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the embodiments. Note that the above-described embodiments and the modified embodiments described below can be used in appropriate combinations, but detailed description thereof will be omitted.

[0055] (Variations) (1) In the present embodiment, the processing of a response to a question has been described as an example, but the processing is not limited to this. The processing can be similarly performed for other documents.

[0056] (2) In the present embodiment, when extracting candidate words with the same context, character strings including candidate words with the same parts of speech as the spelling variation and the words before and after it are extracted. However, this is not limiting. For example, patterns may be created for the parts of speech of the spelling variation and the words before and after it when the spelling variation is used as a candidate word, and character strings from documents in the document storage unit that satisfy each pattern may be extracted and tallied.

[0057] (3) In the present embodiment, the document storage unit stores a corrected response sentence that has been corrected for spelling variations, but the present invention is not limited to this. A simple input document may also be stored in the document storage unit.

[0058] (4) In this embodiment, the display of spelling variations in answer sentences has been described. Here, the display of the spelling variations may be different depending on, for example, the number of each candidate word counted or the proportion of each candidate word in the total number. For example, the higher the number or proportion of usage of a spelling variation in an answer sentence, the darker the color may be.

[0059] (5) In the present embodiment, verification of spelling variations is mainly described as an example, but this is not limiting. Other processes for verifying a document may be performed at the same time.

[0060] (6) In the present embodiment, an example has been described in which a document verification server and a user terminal are included, but this is not limiting. A standalone configuration consisting of only a client terminal having the functionality of a document verification server may also be used.

[0061] (7) In the present embodiment, the document verification server has a storage unit including a spelling variation dictionary, a unified rule storage unit, and a document storage unit, but the present invention is not limited to this. For example, the document storage unit may be stored in an external server, and the document verification server may be connected to the external server that stores the verification information storage unit via a communication network or directly. [Explanation of symbols]

[0062] 1. Document Verification Server 5. User terminal 10 Control Unit 11 Document Reception Department 12 Morphological analysis processing unit 13 Orthographic Variant Identification Section 14 Candidate word extraction part 15 Unified Rules Verification Department 16. Stored Document Search Unit 17 List Generation Unit 18 Result output section 19 Document Update Registration Department 20 Memory section 21a Verification Program 23 Dictionary of spelling variations 24 Unified rule memory 26 Document storage unit 50,80,80-2 Verification screen 54,84 Answer text area after analysis 55,85,85-2 Analysis result area 70 candidate word list 71a, 71b, 71c unified button 72a, 72b, 72c, 72d Edit button 89 Save button 100 Document Verification System

Claims

1. A spelling variation detection device for detecting spelling variations in a document, comprising: a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among words contained in the document to be verified; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the words searched by the document search means, for a character string including a word having the same context as the spelling variation identified by the spelling variation identification means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the number of character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; A spelling variation detection device comprising:

2. A spelling variation detection device for detecting spelling variations in a document, comprising: a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among words contained in the document to be verified; a part-of-speech acquiring means for acquiring the part-of-speech of the spelling variation word identified by the spelling variation identifying means and the part-of-speech of the words before and after the spelling variation word in the document to be verified; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the word searched by the document search means, a string including the word, the parts of speech of which and the words before and after the word match each of the parts of speech acquired by the part-of-speech acquisition means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; A spelling variation detection device comprising:

3. A spelling variation detection device for detecting spelling variations in a document, comprising: a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; a unified rule storage unit that stores unified terms for words; a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among words contained in the document to be verified; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the words searched by the document search means, for a character string including a word having the same context as the spelling variation identified by the spelling variation identification means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; a document output means for outputting the document to be verified and highlighting the spelling variations identified by the spelling variation identifying means; a unified rule confirmation means for confirming whether the spelling variation word identified by the spelling variation identification means matches the word in the unified rule storage unit; a modified word output means for outputting the unified word corresponding to the matched word instead of the output from the candidate word output means when the unified rule confirmation means confirms a match; A spelling variation detection device comprising:

4. In the notation variation detection device according to claim 3, The document output means outputs the spelling variation words that the unified rule checking means has confirmed to match in a highlighted manner different from the output by the candidate word output means.

5. A spelling variation detection device for detecting spelling variations in a document, comprising: a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; a morphological analysis means for performing a morphological analysis on a document to be verified; a word extraction means for extracting the words from the morphological analysis result by the morphological analysis means; a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among the words included in the document to be verified and extracted by the word extraction means; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the words searched by the document search means, for a character string including a word having the same context as the spelling variation identified by the spelling variation identification means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; A spelling variation detection device comprising:

6. In the writing variation detection device according to any one of claims 1 to 5, The spelling variation detection device includes a document registration means for registering the verification target document after the measures regarding the spelling variation have been taken in the document database.

7. A computer that detects variations in the writing of a document a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; The computer a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among words contained in the document to be verified; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the words searched by the document search means, for a character string including a word having the same context as the spelling variation identified by the spelling variation identification means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the number of character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; A program to make it function as such.

8. A computer that detects variations in the writing of a document a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; The computer a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among words contained in the document to be verified; a part-of-speech acquiring means for acquiring the part-of-speech of the spelling variation word identified by the spelling variation identifying means and the part-of-speech of the words before and after the spelling variation word in the document to be verified; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the word searched by the document search means, a string including the word, the parts of speech of which and the words before and after the word match each of the parts of speech acquired by the part-of-speech acquisition means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; A program to make it function as such.

9. A computer that detects variations in the writing of a document a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; a unified rule storage unit that stores unified terms for words; Equipped with The computer a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among words contained in the document to be verified; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the words searched by the document search means, for a character string including a word having the same context as the spelling variation identified by the spelling variation identification means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; a document output means for outputting the document to be verified and highlighting the spelling variations identified by the spelling variation identifying means; a unified rule confirmation means for confirming whether the spelling variation word identified by the spelling variation identification means matches the word in the unified rule storage unit; a modified word output means for outputting the unified word corresponding to the matched word instead of the output from the candidate word output means when the unified rule confirmation means confirms a match; A program to make it function as such.

10. A computer that detects variations in the writing of a document a spelling variation dictionary that stores a plurality of pairs of candidate words for a word associated with one spelling variation; The computer a morphological analysis means for performing a morphological analysis on a document to be verified; a word extraction means for extracting the words from the morphological analysis result by the morphological analysis means; a spelling variation word identification means for identifying, as a spelling variation word, a word that matches one of the candidate words in the spelling variation dictionary among the words included in the document to be verified and extracted by the word extraction means; a candidate word extraction means for extracting a plurality of candidate words included in the set of the spelling variation dictionary that includes the spelling variation word identified by the spelling variation identification means; a document retrieval means for retrieving a word that matches any one of the plurality of candidate words extracted by the candidate word extraction means from a plurality of registered documents stored in a document database; a context search means for searching, from the words searched by the document search means, for a character string including a word having the same context as the spelling variation identified by the spelling variation identification means; a usage counting means for counting the words retrieved by the document search means for each candidate word and for counting the character strings retrieved by the context search means for each candidate word; a candidate word output means for outputting the number of words counted by the usage counting means in association with the candidate words; A program to make it function as such.

Images