JP5205028B2 - Handwritten annotation management device and interface - Google Patents

Description

  The present invention relates to a technique for calculating the importance of handwritten annotations in a document management apparatus including handwritten annotations.

  In recent years, techniques for digitizing handwritten memos on documents have become widespread. One of them is a tablet PC. This is a personal computer equipped with a pen device, and the contents drawn on the display by the pen device can be electronically captured. Another example of the above technique is a digital pen. By reading fine dots printed on paper with a camera built in the pen, handwritten memos on paper can be digitized.

  With the spread of these devices, a technique for managing electronic handwritten memos as document annotations together with document information has also spread. For example, in Microsoft Office 2007 developed by Microsoft Corporation, a handwritten memo can be added to a word processor document or a presentation document and stored using a tablet. Patent Document 1 discloses a technique for managing an electronic document and a handwritten memo on paper together. These technologies have made it possible to electronically store and manage handwritten annotations on documents.

  With these technologies as a background, in recent years, techniques that analyze handwritten annotations on documents and use them for information display and retrieval have attracted attention. In the work of public offices and companies, important parts of a document are recorded by writing underlines and circles. Since such handwritten annotation records often contain important knowledge in business, it is useful to search and browse them electronically. When this is realized, the number of annotations to be managed becomes enormous, so it is important to preferentially present what is important in the search. That is, a means for calculating the importance of each annotation for a search query consisting of a word string is important.

  As a technique for realizing such a search for handwritten annotations, a technique has been proposed in which handwritten annotation information is searched for with importance based on the contents of entries and document contents. For example, Non-Patent Document 1 proposes a method of calculating the importance based on the similarity between the search keyword and the text near the place where the handwritten annotation is entered. Patent Document 2 discloses means for calculating the importance level of handwritten annotations based on the importance level of the word / phrase of the document content of the part where the handwritten annotations are written. Further, Patent Document 3 discloses a means for designating a specific document range and retrieving only handwritten annotations written in the range. By these means, it is possible to retrieve annotation information based on the text content of the annotated document.

JP2005-135211 JP2004-206654 JP2004-199484 JP2006-301695 Gene Golovchinsky, Morgan N .; Price, and Bill N .; Schilit, “From Reading to Retrieval: Freeform Ink Annotations as Queries” Proceedings of ACM SIGIR 99, ACM Press, pp. 19-25, 1999.

  However, underline is often entered in similar sentences in various places in business, and similar sentences are used in the importance calculation means for annotations based on word similarity and importance as described above. The annotations attached to cannot be distinguished well. In business, the correspondence between a part of one document and a part of another document, the degree of strength of the correspondence, and the like may be explicitly recorded by annotation. Such a record expresses the process and knowledge of the worker's work, and if it is used when searching for annotation information, it is expected that the search will be more efficient than the search method based on the conventional text contents. The

  Patent Document 4 discloses a technique for interpreting handwritten additional marks attached to a document and displaying the contents of the document side by side for each type of mark, as an example of utilizing correspondence information based on such annotations. Yes. However, it only relates to a method for displaying handwritten annotations in one document, and is not a technique for efficiently extracting information from a large number of annotation information recorded.

  Therefore, in the present invention, in a document management apparatus including handwritten annotation information, when searching for annotation information, a search result is output using the association information between the annotations by the user. Provided is an annotation retrieval means capable of searching for annotation information by utilizing knowledge of a work process.

  In order to solve the above problems, the present invention provides a document management system, a search system, and a search method as follows.

  A document management system for electronically acquiring handwritten information handwritten on a document, and extracting a range of a target sentence and a mark on the sentence from the acquired handwritten information into an annotation information database Annotation information analyzing means to be registered and the handwritten information obtained in the same business flow as the handwritten information obtained from the handwritten information different from the handwritten information, and from the document in which the mark corresponding to the mark is handwritten A comparison annotation registration unit that extracts the range of the sentence and registers it in the comparison annotation information database as a comparison annotation.

Further, the comparison annotation registration unit acquires evaluation information for the comparison annotation from the handwritten information and registers it in the comparison annotation database. Furthermore, the evaluation that is handwritten as the evaluation information for the comparison annotation It has an evaluation value database in which combinations of symbols and their evaluation values are registered.

  Furthermore, the comparison annotation registration unit extracts a common word and a difference word between ranges of a target sentence in a document in which a corresponding mark is handwritten, and registers the extracted common word and difference word in a comparison annotation information database.

  Furthermore, the common word and the difference word are extracted by extracting feature words in each sentence and comparing the feature words.

  In addition, the search system is handwritten information on the electronically acquired document, the annotation information database in which the range of the target sentence and the mark on the sentence are registered, and the handwritten information is acquired. A comparative annotation in which the range of the target sentence extracted from a document in which the mark corresponding to the mark is handwritten is registered as a comparative annotation, which is different from the handwritten information acquired in the same workflow An information database, a means for receiving input of a keyword, an annotation search unit that searches the annotation information database for a sentence including the input keyword, a sentence searched by the search part, and a contrast annotation corresponding to the sentence Annotation evaluator that extracts registered sentences from the contrast annotation information database and calculates the importance of the contrast annotation; and Interpretation based on the calculated importance by the evaluation unit, characterized and a display means for displaying the handwritten information to the extracted sentence.

  Further, the contrast annotation information database has evaluation information for the contrast annotation acquired from the handwritten information, and the annotation evaluation unit rearranges the sentence sets extracted based on the importance of the evaluation information. The handwritten information is displayed on the display means in the order of the rearrangement.

  Also, a search method using a search keyword input means, search means, display means, and database, wherein the database is handwritten information on an electronically acquired document, and the range of the target sentence and the sentence Annotation information database in which a mark is registered and information different from the handwritten information acquired in the same business flow as the handwritten information is acquired, and the mark corresponding to the mark is handwritten A comparison annotation information database in which a range of a target sentence extracted from a document is registered as a comparison annotation, the input unit accepts an input of a search keyword, and the search unit includes a sentence including the input keyword Is retrieved from the annotation information database, and the contrast between the retrieved sentence and the sentence in which the contrast annotation corresponding to the sentence is registered. Extracted from the interpretation information database to calculate the importance of the contrast annotation, retrieval system the display unit, based on the calculated degree of importance, and displaying the handwritten information to the extracted sentence.

  As a result, not only handwritten annotations with similar text content, but also importance can be assigned to annotations of parts that have been previously determined by the operator to be related to the annotation content. It is possible to search important annotations preferentially by using the working knowledge of.

  Further, in the present invention, the writer intended by managing the business flow and a set of documents included in the business flow and detecting a set of annotations having the same mark written in the documents in the same business flow. An annotation correspondence can be extracted.

  In addition, the present invention increases the importance of not only the high-consistency of the search query and the text but also the annotation linked to the annotation by such a contrast, thereby not only the text content but also the intention of the writer. We propose a method to calculate importance by reflecting it.

  In the present invention, not only the annotation information output as a search result is displayed in a list format, but also the comparison relationship of annotations is displayed in a list format in order of importance, or the comparison relationship of annotations is displayed in a network form. We propose a method for visualizing the relationship between annotations written by past writers.

  According to the present invention, a set of document contents that are explicitly described as having a relationship with a writer is extracted from a large amount of document information accumulated in a company, and the relationship is not used to depend on the text contents. The important document information can be preferentially searched by the scale.

  The configuration of the information management apparatus according to the embodiment of the present invention will be described with reference to FIG. An information management apparatus as an embodiment of the present invention includes an information management server 101 and an information management client 102.

  The information management server 101 includes a user information DB 103 for storing information on a user who uses the apparatus, a document information DB 104 for storing document information, a stroke information DB 105 for storing handwriting added to the document, and analyzing strokes and document contents. Annotation information DB 106 that stores the shape of a stroke that is underlined or circled and an annotation content that indicates which part is underlined, and workflow information that stores information about the workflow and the documents used therein DB 107, contrast annotation information DB 108 that stores correspondences between annotations that are considered to be intended by the writer, and evaluations that store correspondences between evaluation symbols and evaluation values when the writer explicitly evaluates the annotations Value DB 130 receives entry data from client and stores data in stroke information DB and annotation information DB Fill data registration unit 109, and a annotation information search processing unit 110 returns the list of receiving annotation information retrieval query from the client. The entry data registration unit 109 is a stroke information registration unit 120 that stores information in the stroke information DB, and an annotation analysis that analyzes a handwritten shape such as an underline or a circle and a target sentence and stores the information in the annotation information DB. The registration unit 121 includes a comparison annotation registration unit 122 that extracts a correspondence between annotations based on an additional mark and stores information in the comparison annotation information DB. The annotation information search processing unit 110 calculates the importance of the annotation information with respect to the search query based on the content of the annotation search unit 123 using keywords for acquiring the annotation information including the text of the search query as the content, the annotation information and the contrast annotation information. An annotation evaluation unit 124 that sorts annotation information based on importance and outputs a search result to a client as a search result.

  The information management client 102 is a terminal for each user to digitize handwritten annotations on a document or to search and browse annotation information by using the information management client 102. The information management client 102 reads entry data from a pen device and transmits it to a server. The data reading unit 111 includes an annotation information search unit 112 that receives a search query from a user, transmits it to the server, and displays the results returned from the server in a list format on the screen. A pen device 113 is connected to the client and can read an electronic document entry.

  First, FIGS. 2 to 5 show information management formats in the information management server.

  The management format of the user information DB 103 will be described with reference to FIG. The user information includes a user ID 201 and a user name 202 that are uniquely assigned to each user.

  The management format of the document information DB 104 will be described with reference to FIG. The document information includes a document ID 301 uniquely assigned to each document, a creation user ID 302 representing a user who created the document, a creation date and time 303, an update date and time 304, electronic document data 305 storing an electronic file of the document, and the number of pages of the document 306, a document instance ID list 307. The document instance ID is intermediate data to be filled in when filling in the document. When different users write in the same document, it is managed to write the same document while dividing it. Is. For example, when different users want to fill in a document, separate document instances are generated and written.

  The management format of the stroke information DB 105 will be described with reference to FIG. The stroke information is composed of a stroke ID 401 uniquely assigned to each stroke, a document instance ID 402 in which the stroke is entered, a page number 403 in which the stroke is entered, a user ID 404 in which the stroke is entered, an entry start time 405, and a sampling point 406. And has a pointer 407 to a list of sampling points constituting the stroke. The sampling point is composed of an X coordinate 408, a Y coordinate 409, and a difference time 410. The stroke information not only holds the shape of the stroke by the list of sampling points, but also holds information such as the writer and entry time.

  The management format of the annotation information DB 106 will be described with reference to FIG. The annotation information includes an annotation ID 501 that is uniquely assigned to each annotation, an ID 502 of the user who entered the annotation, a list 503 of stroke IDs that make up the annotation, and a part of the annotation that is considered to be assigned by the writer. A mark 504, an annotation shape 505 such as an underline or a circle, and an annotated character string 506 that stores a text to be annotated such as an underlined text or a circled text. 510 is an example of an annotation. In 510, the range of the document content is selected by an underline, and the symbol “1a” is entered in the upper left. This apparatus detects the underline and the mark “1a” from such a stroke and registers it as annotation information.

  The management format of the business flow information DB 107 will be described with reference to FIG. The business flow information includes a business flow ID 601 that is uniquely assigned to each business flow, a business flow start date and time 602 and an end date and time 603, and a set of document information 604 included in the business flow. The document information 604 included in the business flow includes a corresponding document instance ID 605, a corresponding business flow ID 606, an ID 606 of a parent document instance that is a document created or used prior to the document in the business flow, and It consists of an ID 607 of a child document instance which is a document created or used after the document. A set of document information in the business flow has a network structure as shown in FIG.

  The management format of the contrast annotation information DB 108 will be described with reference to FIG. The comparison annotation information includes a comparison annotation ID 701 that is uniquely assigned to each detected comparison annotation, a set of IDs 702 of annotations constituting the comparison annotation, an evaluation value 703 of the comparison annotation explicitly indicated by the writer, A common word list 704 between two commented character strings and a list of difference words 705 and 706 between two commented character strings. A set of the annotation 710 and the annotation 711 is an example of a contrast annotation. The range of the annotation 710 is selected by vertical brackets, and a mark “1a” is written at the upper left. The range of the annotation 711 is underlined, and a mark “1a” is written in the upper left. Furthermore, document instances with these annotations belong to the same business flow. At this time, this apparatus detects that the marks of both annotations are the same, and registers the set as a comparison annotation in the comparison annotation DB.

  Next, information registration processing when a user enters an annotation on a document with a pen will be described with reference to FIGS. First, the overall flow of the information registration process will be described with reference to FIG. When the user 807 fills in the document instance 809 using the pen device 808, the entry data reading unit of the client 801 acquires stroke coordinates and time information from the pen device, generates stroke information 802, and transmits it to the server 803. To do. At this time, a pen tablet, a tablet PC, or the like may be used as the pen device, or a pen device that directly fills in paper such as an Anoto digital pen may be used. When the server receives the transmitted stroke information, the stroke information registration unit 804 registers the stroke information in the stroke information DB, then the annotation analysis / registration unit 805 analyzes the shape of the stroke and stores the result in the annotation information DB. sign up. Finally, the comparison annotation registration unit 806 acquires the annotations with the same mark among the annotations entered in the document instances in the same business flow and registers them in the comparison annotation information DB. As a result, each time the user enters an annotation, information such as which part is underlined and the correspondence relationship of the annotation explicitly indicated by the user are automatically registered in the server.

  A flowchart of the annotation analysis / registration unit in the information registration process is shown in FIG. The annotation analysis / registration unit receives the stroke information transmitted from the client in step 901, and recognizes the shape of the stroke such as a line or a circle in step 902. In step 902, if a preset annotation shape such as an underline or a circle is recognized, the process proceeds to step 904 by a conditional branch in step 903. If the annotation shape is not recognized, the annotation analysis / registration unit ends. In step 904, the stroke shape is compared with the position information of the document content, and the range covered by the annotation is detected. For example, in the case of an underline, it is detected which sentence is underlined from the coordinates of the underline and the coordinates of the document text. In step 905, the symbol entered as the annotation identifier is detected as a mark. In step 906, the annotation shape, annotation range, and mark information are registered in the annotation information DB as annotation information.

  FIG. 10 shows a flowchart of the contrast annotation registration unit in the information registration process. The contrast annotation registration unit acquires the annotation information 1 registered in the annotation analysis / registration unit in step 1001, acquires the information of the document instance in which the annotation information 1 is entered in step 1002, and the business to which the document instance belongs. Get flow information. Next, in step 1003, all annotation information entered in all document instances belonging to the business flow is acquired, and in step 1004, all annotations whose marks are the same as the annotation information 1 are acquired. . In step 1005, a set of the annotation information 1 and the annotation information acquired in step 1004 is registered in the comparison annotation information DB as a comparison annotation. In step 1006, the evaluation value of the contrast annotation is set from the evaluation information for the contrast annotation entered by the user. The evaluation value entry method here can be performed by, for example, entering a mark symbol and an evaluation symbol side by side in the margin. Here, as shown in FIG. 21, the evaluation value represented by the evaluation symbol may be registered in advance in the evaluation value DB together with the evaluation symbol. Finally, in step 1007, common / difference words of the two annotations in the contrast annotation are extracted by text processing. Here, the common / difference word extraction can be realized, for example, by first extracting feature words from the character string to be annotated by general text processing, and checking whether or not they appear in common in two annotations.

  An example of the evaluation symbol entry method in step 1006 of the contrast annotation registration unit is shown in FIG. In this example, an annotation 1103 is entered in the document instance 1101 and an annotation 1104 is entered in the document instance 1102, which have the same mark “1a” and are registered as contrast annotations. Further, the evaluation information 1105 of the contrast annotation is entered in the margin of the document instance 1102. At this time, the apparatus detects such a description by character recognition, and considers that the mark “1a” is evaluated as “x”. Here, a value assigned in advance to the evaluation symbol x is set as the evaluation value of the comparison annotation. In the example of FIG. 11, an evaluation value corresponding to the evaluation symbol y is set for the contrast annotation having the mark “2b”. Here, evaluation symbols are collectively written for each type of mark in the margin of the document. However, a method may be adopted in which evaluation symbols are individually written on the side of each annotation. An evaluation value is registered according to the entered symbol. When registered, the evaluation value may be registered numerically as shown in FIG. By registering the evaluation value of the annotation by the user's entry in this way, the importance of the annotation can be calculated more reflecting the intention of the writer when the annotation is searched.

  The evaluation for the above comparison annotation can be changed after the evaluation value is registered in the database. FIG. 12 shows an example of evaluation change by adding to a document instance. In this example, the evaluation symbol previously entered in the margin is hatched, and a new evaluation symbol is entered thereon. The apparatus detects the oblique lines and the evaluation symbols by character recognition, and changes and registers the evaluation value of the contrast annotation for the corresponding mark in the database. The evaluation change may be performed not only by filling in the document instance but also by keyboard input from a PC screen or the like. An example is shown in FIG. In this example, the document instances belonging to the business flow on the screen 1301 and the marks entered therein are displayed as a table. In each column of the table, an evaluation symbol for the corresponding mark entered in the corresponding document instance is displayed. For example, the document D0032154 indicates that the evaluation symbol X is entered for the item 1a, and the evaluation symbol Y is entered for the items 2b, 3c, and 4d. Further, not all the evaluation symbols for each item are entered for each document. For example, since the evaluation symbol for the item 1a is not entered in the document D0032162, the corresponding column is blank. The user can change the value of each column from the screen of the PC. In this case, the apparatus corrects the evaluation value of the corresponding contrast annotation registered in the contrast annotation information DB. In the example of FIG. 13, the evaluation symbol for the item 1a in the document D0032154 is corrected from X to Y, and the importance of the corresponding contrast annotation information is changed from the evaluation value 2.0 assigned to the evaluation symbol X accordingly. The evaluation value assigned to Y has been changed to 1.0. FIG. 14 shows a flowchart of the common / difference word extraction unit in the contrast annotation registration step 1006. The common / difference word extraction unit acquires the target annotation information 1 and 2 in step 1401, and in step 1402 extracts the feature word from the annotation target character string of each annotation information. This feature word extraction can be realized using a dictionary-based method or a method based on word appearance frequency such as TF-IDF. Next, in each of the feature words extracted in step 1403, a word common to both annotations is registered in the comparison annotation information as a common word. In step 1404, a characteristic word in the character string to be annotated in the annotation information 1 and not appearing in the character string to be annotated in the annotation information 2 is registered in the contrast annotation information as a difference word 1. Similarly, in step 1405, the annotation information A feature word in the character string to be annotated 2 that does not appear in the character string to be annotated in the annotation information 1 is registered in the contrast annotation information as a difference word 2. These common words and difference words are key in text search, but are also useful for characterizing the relationship between two annotations in contrast annotations.

The above is the process flow of the handwritten information registration unit. As a result of the handwritten information registration unit, annotation information indicating what kind of document is underlined or circled, and contrast annotation information indicating which part the writer considers to be relevant are obtained. It can be expected that such information provides an effect that cannot be obtained by simply comparing the similarity of simple texts in a search. An example is shown in FIG.
Looking at the contents of the three document parts 1501, 1502, and 1503, although these contents are highly similar to each other, 1501 and 1502 include the common words “feature word” and “extracted”, so the similarity is high. Although it is considered high, 1503 does not include a common word, so it is considered that the similarity between these two documents is low. Therefore, for example, when searching with the keyword “feature word”, 1501 and 1502 can be acquired, but 1503 cannot be acquired. On the other hand, if a correspondence between documents is found from handwritten annotations as in the present invention, if someone has written an annotation that is related to 1504, 1505, 1506, the similarity of text contents The relationship between these documents can be obtained regardless. Since such handwritten annotations are entered by the writer, it is highly likely that they are actually related. In such a state, for example, when searching with the keyword “characteristic word”, not only 1504 and 1505 but also the document 1506 can be acquired based on this correspondence. As described above, even if there is a notation fluctuation or a phrase difference by the method according to the present invention, an effect that the search can be made efficient by utilizing the past thoughts / knowledge of the worker can be obtained. Next, a process for searching for annotation information managed by the server by the user will be described with reference to FIGS.

  First, the overall flow of the search process will be described with reference to FIG. When the user inputs a search query 1601 to the annotation search unit of the client, the annotation search unit transmits the search query to the server. The search query is composed of a word string as a keyword. When the server receives the search query, the annotation search unit 1602 by keyword first acquires all the annotation information that includes the word included in the word string of the search query in the character string to be annotated. Next, the annotation evaluation unit 1603 calculates the importance of each annotation and the importance of the annotation corresponding to these annotations. Finally, the search result generation unit 1604 rearranges the annotation information obtained by the keyword-based annotation search unit 1602 in the order of importance obtained by the annotation evaluation unit 1603, and sends the top N list 1605 to the client as a search result. Send. When the client receives the search result, the client displays a list of annotation information including the shape of the annotation and the content of the document. Here, the annotation evaluation unit calculates importance by using information other than text information such as user evaluation and time information, and can acquire information related to relevance that cannot be obtained by similarity based on keywords.

  A flowchart of the annotation evaluation unit in the search process is shown in FIG. The annotation evaluation unit receives the annotation information for which importance is calculated in step 1701, and calculates the importance of the annotation information in step 1702. Here, the importance of the annotation information may be calculated based on the similarity between the search keyword and the character string to be annotated, and when the comparison annotation is to be searched preferentially, the comparison annotation information including the annotation information is included. You may make it raise the importance of what has many numbers. Also, the searcher may specify the person and date and time conditions so that the importance of the writer who fills in those conditions and the annotation information having the date and time of entry may be increased. It is also possible to search for items preferentially. As an example of a specific importance calculation method, a value obtained by calculating the similarity between a search keyword and a character string to be annotated by a cosine distance is d, the number of contrast annotation information including the annotation information is n, the annotation information The time of entry is t1, the date and time specified as the search condition is t2, the user ID of the person who entered the annotation information is u1, and the user ID specified as the search condition is u2, and the importance of the annotation information is set appropriately. The coefficients a1 to a4 can be determined and calculated using the formula: a1 × d + a2 × n + a3 × | t1−t2 | + a4 × eq (u1, u2). Here, the function eq returns 1 if the two arguments are equal, and 0 otherwise. In the importance calculation, in addition to the above method, the similarity between the search keyword and the character string to be annotated can be obtained by a method other than the cosine distance, and the importance calculation formula is not a simple addition but a quadratic function or A more complicated calculation formula such as a square root can also be used. Next, in step 1703, all the comparison annotation information including the annotation information is acquired. Finally, in step 1704, the importance levels of all the annotation information included in the comparison annotation information acquired in step 1703 are calculated. Here, the importance of the annotation information is calculated based on the importance calculated in step 1702, but an evaluation value of a comparison annotation may be further added. Moreover, you may reflect the characteristic of the common word / difference word of contrast annotation in importance calculation. For example, when the search keyword specified by the searcher is not included in the common words of the contrast annotation, the importance may be increased, and even if the writer determines that the search keyword specified by the searcher is closely related to the content of the search keyword Regardless, since the search keyword is not directly included, it is possible to increase the importance of the portion whose importance is lowered in step 1702. An example of a specific method of such calculation is as follows. The importance of the annotation information calculated in step 1702 is v, the evaluation value of the comparison annotation is e, and the common word between the search keyword and the common word of the comparison annotation is shown. The value indicating whether or not a word is included is c (c = 0 if a common word is included, c = 1 otherwise), and the importance of the annotation information is determined by appropriate coefficients b1 to b3. B1 × v + b2 × e + b3 × c. Also, the importance may be increased when the amount of difference words in the contrast annotation is large, and this also has the effect of increasing the importance of the portion that becomes less important in the simple text search in Step 1702. It is done. As a result, not only the annotation caught in the text search but also the annotation that the writer has determined to be related to them can be included in the search result.

  Finally, an example of the output form of the search result is shown in FIG. Here, the document contents near the annotation are displayed in a list in order of the importance of the annotation. Furthermore, when there are contrast annotations, the associated annotations are also displayed side by side, so that it is easy to understand the purpose of the annotation. In this example, the search is performed using the search keywords “aluminum” and “mixed”. When this search keyword is received, annotation information 1804, 1805, 1806 is extracted by keyword search in the annotation search unit 1602 of the server, and its importance is calculated in step 1702 of the annotation evaluation unit. Further, annotation information 1807 and 1808 related to them by contrast annotation are extracted by steps 1703 and 1704 of the annotation evaluator, and their importance is calculated. Then, in the search result generation unit 1604, if there are annotation information related by the comparison annotation, they are paired, and the extracted annotation information is rearranged in descending order of importance and output. Looking at the search results, although the annotation information 1807 does not include the words “aluminum” and “mixed”, it is determined that the content is related and the search is possible.

  The fact that a search resistant to the fluctuation of the notation can be realized by this search function will be described with reference to the example of FIG. FIG. 19 shows the result of executing an annotation search with the search keyword “personal computer”. The two search results are annotations that each contain the word “personal computer”, but the annotations that are related to each other by contrast annotations include the words “personal computer”, “word processor”, and “personal computer”. What is not displayed. Of course, these words are synonyms and highly related, but the conventional text search could not be output as a search result. In this method, if there is a person who has paid attention to these relationships in the past, these relationships can be acquired.

  FIG. 20 shows an example of a search result output in which the search result is displayed not in a list but in a graph structure in which a document is a node and a contrast annotation relation is an edge. Here, based on the evaluation value input by the user, when the similarity is high, the edge may be shortened and the distance between the nodes may be shortened. When the annotation is corrected as shown in FIGS. 12 and 13, the display of linking according to the importance may be changed to a different display. Such a display method allows a more concise overview of the relationship between annotations.

It is a lineblock diagram of information management device 10 which is an embodiment of the present invention. This is an information storage format of the user information DB 103 in the information management apparatus 10. This is an information storage format of the document information DB 104 in the information management apparatus 10. This is the information storage format of the stroke information DB 105 in the information management apparatus 10. This is an information storage format of the annotation information DB 106 in the information management apparatus 10. This is the information storage format of the business flow information DB 107 in the information management apparatus 10. This is an information storage format of the comparison annotation information DB 108 in the information management apparatus 10. 3 is a diagram showing an overall flow of entry data registration processing in the information management apparatus 10; FIG. It is a flowchart which shows the flow of a process of the annotation analysis and registration part in an entry data registration process. It is a flowchart which shows the flow of a process of the contrast annotation registration part in an entry data registration process. It is a figure which shows the example of entry of the evaluation symbol with respect to contrast annotation information. It is a figure which shows the example which corrects the evaluation symbol with respect to contrast annotation information by handwriting. It is a figure which shows the example which corrects the evaluation symbol with respect to contrast annotation information from a PC screen. It is a flowchart which shows the flow of a process of the common / difference word extraction part in a contrast annotation registration part. It is a figure explaining the effect by managing contrast annotation information. FIG. 4 is a diagram showing an overall flow of annotation information search processing in the information management apparatus 10. It is a flowchart which shows the flow of a process of the annotation evaluation part in an annotation information search process. It is a figure which shows the example output by the list of the search result in an annotation information search process. It is a figure explaining the effect with respect to the annotation information search by managing a contrast annotation. It is a figure which shows the example of an output by the graph structure of the search result in an annotation information search process. This is an information storage format of the evaluation value DB 130 of the information management apparatus 10.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Information management system, 101 ... Information management server, 102 ... Information management client, 103 ... User information DB, 104 ... Document information DB, 105 ... Stroke information DB, 106 ... Annotation information DB, 107 ... Annotation registration part, 108 ... Annotation search processing unit 109... Annotation search unit 110... Entry data reading unit 111.

Claims (14)

Means for electronically acquiring handwritten information handwritten on a document;
An annotation information analysis means for extracting a range of a target sentence and a mark to the sentence from the acquired handwritten information and registering it in an annotation information database;
It is handwritten information different from the handwritten information acquired in the same business flow as the handwritten information is acquired, and extracts a range of a target sentence from a document in which a mark corresponding to the mark is handwritten, A comparison annotation registration unit for registering in the comparison annotation information database as a comparison annotation;
The comparison annotation registration unit extracts a common word and a difference word from a range of target sentences in a document in which a corresponding mark is handwritten, and registers it in a comparison annotation information database. .   The document management system according to claim 1, wherein the comparison annotation registration unit acquires evaluation information for the comparison annotation from the handwritten information, and registers the evaluation information in the comparison annotation database. .   3. The document management system according to claim 2, further comprising an evaluation value database in which a combination of handwritten evaluation symbols and evaluation values is registered as evaluation information for the comparison annotation. system.   The document management system according to claim 3, wherein the common word and the difference word are extracted by extracting feature words in each sentence and comparing the feature words. system.   3. The document management system according to claim 2, wherein the comparison annotation registration unit registers the change in the comparison annotation database when the evaluation information entered for the document is changed. Document management system.   3. The document management system according to claim 2, further comprising an interface for changing the evaluation information registered in the contrast annotation database. Annotation information database that is handwritten information on an electronically acquired document, in which the range of the target sentence and the mark on the sentence are registered,
The range of the target sentence extracted from the document in which the mark corresponding to the mark is handwritten and is different from the handwritten information acquired in the same business flow from which the handwritten information is acquired is compared. A contrast annotation information database registered as an annotation;
Means for accepting keyword input;
An annotation search unit that searches the annotation information database for a sentence including the input keyword;
An annotation evaluation unit that extracts a sentence searched by the search unit and a sentence in which a comparison annotation corresponding to the sentence is registered from the comparison annotation information database, and calculates the importance of the comparison annotation;
Display means for displaying handwritten information on the extracted text based on the importance calculated by the annotation evaluation unit,
In the comparison annotation information database, a common word and a difference word extracted between ranges of a target sentence in a document in which a corresponding mark is handwritten are registered as a comparison annotation.   The search system according to claim 7, wherein the importance is a similarity between the keyword and a character string in a range of the sentence.   The search system according to claim 7, wherein the importance is a large amount of information of the comparison annotation.   The search system according to claim 7, wherein the importance is based on at least one of a handwritten input person and a date and time of the business flow.   8. The search system according to claim 7, wherein the comparison annotation information database has evaluation information for the comparison annotation acquired from the handwritten information, and the annotation evaluation unit extracts based on importance of the evaluation information. A search system, wherein the set of written sentences is rearranged, and the handwritten information is displayed on the display means in the order of the rearrangement. The search system according to claim 7,
The importance is the value of whether or not the keyword is included in the common word or the number of different words.   8. The search system according to claim 7, wherein the comparison annotation information database has evaluation information for the comparison annotation acquired from the handwritten information, and the display means includes a plurality of the handwritten information based on the evaluation information. A search system characterized by arranging two sets in a two-dimensional manner. Search keyword input means, search means, display means, a search method using a database,
The database is
Annotation information database that is handwritten information on an electronically acquired document, in which the range of the target sentence and the mark on the sentence are registered,
The range of the target sentence extracted from the document in which the mark corresponding to the mark is handwritten and is different from the handwritten information acquired in the same business flow from which the handwritten information is acquired is compared. A comparative annotation information database registered as annotations,
The input means accepts input of a search keyword,
The search means
A sentence including the input keyword is searched from the annotation information database, and the searched sentence and a sentence in which a comparison annotation corresponding to the sentence is registered are extracted from the comparison annotation information database. Calculate importance,
The display unit displays handwritten information on the extracted sentence based on the calculated importance,
The search method according to claim 1, wherein the comparison annotation information database registers, as a comparison annotation, a common word and a difference word extracted between ranges of a target sentence in a document in which a corresponding mark is handwritten.

Images