JP2004021605A - Information sorting device, method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information sorting device and method and program for easily and accurately sorting information into groups, and for informing a user the groups of the information with sounds. <P>SOLUTION: The information sorting device is provided with a sorting means for sorting information into groups based on a keyword dictionary, a dictionary updating means for updating the keyword dictionary based on configuring elements extracted from the information, and a sound reproducing means for reproducing sound according to the groups obtained by sorting the information. In this case, the keyword dictionary is updated based on the configuring elements extracted from the information by the dictionary updating means (S315). Thus, it is possible to easily and accurately sort the information into groups. Also, a sound reproducing means reproduces a sound according to the sorted groups. Thus, it is possible to inform the user of the groups of the information with those sounds. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
[Prior art]
The present invention relates to an information classification device, a method, and a program.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, a technique for classifying information using a keyword registered in advance has been widely known. The information to be classified covers a wide range such as e-mails, texts, images, and voices distributed via the Internet. By applying such information classification technology, it is possible to extract important information from users in a short period of time from a large amount of information, or to extract important information from users when receiving information as needed through a communication line. You can immediately know that when you receive.
[0003]
By notifying the user of the information with the sound of the classification result, there is a high possibility that the classification result can be immediately notified regardless of whether the user pays attention to the information. In addition, if the classification result is notified by sound, the user can classify the information by using the auditory sense while grasping the information by using the visual sense, thereby facilitating the grasp of the information.
[0004]
Japanese Patent Application Laid-Open No. 2001-282635 discloses a communication device that outputs a melody registered in advance in association with a keyword when a keyword registered in advance is present in an object in the e-mail, when the e-mail is received. Is disclosed. According to this communication device, the operator can immediately know the arrival of an important electronic mail by registering the keyword in advance.
[0005]
[Problems to be solved by the invention]
However, the communication device disclosed in JP-A-2001-282635 has the following problem. First, it is difficult and troublesome to register an appropriate keyword. In order to distinguish important information from unimportant information for an information user, an appropriate keyword must be registered in advance. However, in order to accurately classify important information and unimportant information, it is necessary to register a large number of keywords together with appropriate combination conditions. Second, there is a temporal change in the importance of information. What is important to the user of the information changes over time. Therefore, in order to distinguish important information from information that is not important at one time, keywords must be constantly updated.
[0006]
The present invention has been made in order to solve these problems, and provides an information classification apparatus, method, and program for easily and accurately classifying information into groups and notifying a user of the information groups by sound. The purpose is to provide.
[0007]
[Means for Solving the Problems]
In order to achieve the above object, an information classification device according to the present invention includes: a classification unit configured to classify information into groups based on a keyword dictionary; and a dictionary update unit configured to update the keyword dictionary based on components extracted from the information. Sound reproducing means for reproducing a sound in accordance with the group into which the information is classified.
[0008]
The information used by the dictionary updating means to update the keyword dictionary is not information input for the purpose of direct classification, but specifically, for example, a document created by text input and received through a communication line. Document. That is, the components used by the dictionary updating means to update the keyword dictionary specifically correspond to, for example, relatively small linguistic units (titles, phrases, phrases, words, etc.) input when creating an e-mail. The text is a text corresponding to a relatively small language unit in the received electronic mail, and is a text corresponding to a relatively small language unit included in an HTML file received via the Internet.
[0009]
According to the information classification device according to the present invention, since the dictionary updating unit updates the keyword dictionary based on the components extracted from the information, the information can be easily and accurately classified into groups, and the sound reproducing unit is classified. Since the sound is reproduced according to the group, the group of information can be notified to the user by sound.
[0010]
Further, the dictionary updating means of the information classification device according to the present invention is characterized in that, when information is created by text input, a component of the information is extracted for each section of the text input. In general, keywords registered in a keyword dictionary must be common nouns, proper nouns, and the like whose appearance frequency increases in a specific context. Therefore, in order to extract an appropriate keyword from the text information, a process of decomposing a sentence into words is required. On the other hand, at the time of text input, it is often the case that a character type is converted into a word, a phrase, a phrase, or the like into, for example, a kanji kana mixed sentence, or a word is separated by a space. Therefore, the process of extracting an appropriate keyword from the information to update the keyword dictionary is simplified by the dictionary updating unit extracting the information components for each section of the text input.
[0011]
Furthermore, the dictionary updating unit extracts the components of the information from the received information, so that the information can be accurately classified into groups. This is because, in general, a user of information actively collects information that is important to him / her, so that important information is received more frequently than non-important information.
[0012]
Further, the dictionary updating unit extracts from the information a component having a strong co-occurrence relationship with the keyword registered in the keyword dictionary, and registers the component in the keyword dictionary as a paired keyword and a paired keyword in the keyword dictionary. Classification can be made more accurately. In addition, "pair" means being associated with each other.
[0013]
Further, the sound reproducing means reproduces a sound corresponding to a group to which the information is classified when the information is received, so that when the important information is received, the information can be used immediately. .
[0014]
The information classification device according to the present invention further includes weighting means for registering a weight for the keyword in the keyword dictionary, wherein the classification means classifies the information into groups based on the weight registered for the keyword. . By using weighting, information can be classified more accurately.
[0015]
Further, the sound reproducing means changes the control information for reproducing the sound in accordance with the frequency of appearance of the keyword registered in the keyword dictionary and the frequency of appearance of the information in the displayed portion, whereby the user For example, the user can easily know that the important part of the information is displayed during the scroll display, so that the user can efficiently use the important part of the information. Here, the displayed portion of the information refers to a displayed portion of the information that cannot be entirely displayed on one screen.
[0016]
Furthermore, the information classification device according to the present invention further includes a mapping unit that displays a position corresponding to the correlation between the information and the group on a coordinate plane, so that the tendency of the information correlated with the plurality of groups can be accurately determined. I can tell the user.
[0017]
Further, the dictionary updating means directly registers the intention of the operator by additionally registering the components of the information classified by the manual classifying means in the keyword dictionary as a keyword paired with the group designated by the operator. Can be reflected.
[0018]
Each function of the plurality of means provided in the information classification device according to the present invention is realized by an arbitrary combination of hardware resources whose functions are specified by the configuration itself and hardware resources whose functions are specified by a program. . The functions of the plurality of means are not limited to those realized by hardware resources which are physically independent of each other.
[0019]
Further, the present invention can be specified not only as an invention of an apparatus, but also as an invention of a program, an invention of a recording medium on which the program is recorded, and an invention of a method.
[0020]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 2 is a diagram illustrating a state in which the information classification device 1 according to one embodiment of the present invention is connected to a communication network N and an electronic mail delivery network M. The information classifying device 1 is configured as a personal computer, a personal digital assistant (PDA), a mobile phone, or the like, receives an HTML file from a WWW (World Wide Web) server 2 through a communication network N such as the Internet, and receives electronic files such as a telephone line. The e-mail is transmitted to and received from the e-mail device 3 via the mail delivery network M. Note that the communication network N and the e-mail delivery network M may be the same.
[0021]
FIG. 3 is a block diagram illustrating a hardware configuration of the information classification device 1. 1, the information classification device 1 includes a CPU 11, a ROM 12, a RAM 13, an operation device 14, a communication unit 15, a sound control unit 16, a speaker 17, a display control unit 18, a display device 19, and an external storage device 20.
[0022]
The CPU 11 executes a program stored in the ROM 12 to control each unit of the information classification device 1, executes a processing program, and stores a keyword dictionary W based on components extracted from information such as an HTML file and an electronic mail. Processing for updating, processing for classifying information into groups based on the keyword dictionary W, processing for reproducing sound according to the group to which the information is classified, and displaying the position corresponding to the correlation between the information and the group on a coordinate plane And a process of registering a weight for a keyword in the keyword dictionary W based on a component of information.
[0023]
The ROM 12 is a memory in which control programs and data, a processing program, an e-mail program, a Web browser, and the like necessary for the operation of the CPU 11 are stored in advance, and the RAM 13 temporarily stores programs, various data, and the like. Memory. These programs and various data may be downloaded via the communication unit 15 and stored in the RAM 13 or a predetermined area of the external storage device 20. These programs and various data may be read from a computer-readable storage medium such as a compact disk (not shown) and stored in the RAM 13 or a predetermined area of the external storage device 20.
[0024]
The operating device 14 is, for example, a keyboard or mouse in the case of a personal computer, or a dial button in the case of a mobile phone, and is used by the operator of the information classification device 1 to input various instructions and text.
[0025]
The communication unit 15 is a so-called network interface card, a modem, or the like, and is configured to be connectable to the communication network N and the electronic mail delivery network M.
The sound control unit 16 generates an acoustic signal based on a sound file in which control information for reproducing a sound is described. Then, this acoustic signal is output to the speaker 17 and reproduced. The sound control unit 16 can adjust the volume of the speaker 17, that is, the reproduction volume. When the CPU 11 outputs control information indicating a set value of the reproduction volume, the sound control unit 16 outputs the control information. Adjust the playback volume based on that.
[0026]
The display control unit 18 transmits information such as an HTML file received through the communication network N and an e-mail received through the e-mail delivery network M to a display device including a liquid crystal display panel (LCD) and a CRT under the control of the CPU 11. 19 is output.
[0027]
The external storage device 20 includes a hard disk, a flash memory, and the like, and stores a keyword dictionary, a co-occurrence keyword dictionary, a kana-kanji conversion dictionary, a keyword candidate dictionary, a group table, a melody table, a sound file, and the like, which will be described later. The data format of the sound file may be digitally encoded data such as the MIDI format, or may be data based on a waveform sample data system such as PCM, DPCM, ADPCM, or the like.
[0028]
FIG. 4 is a data flow diagram of the information classification device 1. The processing program generates a communication process 31, a classification process 32, a melody reproduction process 33, a mapping process 34, a morphological analysis process 35, a kana-kanji conversion process 36, a filter process 37, and a dictionary update process 38.
[0029]
The communication process 31 receives an HTML file from the WWW server 2 according to a predetermined protocol, and transmits and receives an e-mail with the e-mail device 3. When the communication process 31 receives information such as an HTML file and an e-mail, it outputs them to the morphological analysis process 35 and the classification process 32.
[0030]
The classification process 32 classifies information output from the communication process 31 into groups based on the keyword dictionary W, and outputs a group number, a group evaluation value, and the like for each information to the melody reproduction process 33 and the mapping process 34.
[0031]
FIG. 5 is a diagram illustrating an example of the keyword dictionary W. The record of the keyword dictionary W includes fields of a group number, a keyword, a setting value, and a song number. The group numbers and keywords included in the same record are managed in the keyword dictionary W in pairs. The group number is assigned to each group in a group table G described later, and the group to which the keyword belongs is determined by the group number. Therefore, registering a group number and a keyword in pairs in the keyword dictionary W is substantially equivalent to registering a group and a keyword in pairs in the keyword dictionary W. Words that appear particularly frequently in a specific context are registered as keywords. The set value indicates the weight for the keyword, and the received information is classified into groups based on the keyword and the weight indicated by the set value. The song number is assigned to each sound file name in a melody table M described later. The setting of the song number for the keyword is optional, and if not set, the sound file to be reproduced is determined by the song number set for the group to which the keyword belongs. The song number is set so that, for example, when the frequency of occurrence of a specific keyword for certain information is extremely high, the sound file set for the keyword is output in preference to the sound file set for the group to which the keyword belongs. It can be used in such cases.
[0032]
FIG. 6 is a diagram illustrating an example of the group table G. The group table G is a table for storing group names for classifying information. The record of this table is composed of four fields: a group number, a group name, a music number, and set coordinates. When the information is classified into groups, a sound file corresponding to the music number of the classified group is played. “Preference” in the group name is a group for classifying keywords representing the preference of the operator, and the method of registering keywords belonging to the group “preference” is different from other groups. Details will be described later. The set coordinates are values for calculating the position of the mark when displaying the mark at a position on the coordinate plane corresponding to the correlation between the information and the group, and will be described in detail later.
[0033]
The classification process 32 classifies information using the above-described keyword dictionary W and group table G, for example, as follows. The classification process 32 extracts text information such as a title, a text, and a file name from the information to be classified, calculates an evaluation value of each group for the information to be classified by the following formula, and divides the information into a group evaluation value. Into the largest group.
Group evaluation value = Σ (number of appearances of keywords belonging to the group × setting value of the keywords)
Here, the set value of the keyword is the “set value” stored in the keyword dictionary W. The number of appearances of a keyword is a value obtained by counting the number of keywords in information to be classified. The classification process 32 can classify not only text information but also image information, sound information, and the like using a file name or the like.
[0034]
For example, assume that the text information has the following contents.
(Information): “I bought a motorcycle model”
When the keyword dictionary W shown in FIG. 5 is used, the group evaluation value of each group is calculated as follows.
Group "Hobbies": 5 ("bike" x 1) + 2 ("plamo" x 1) = 7
Group "work": 0
Group "Taste": 0
Therefore, the text information “bought a motorcycle model” is classified into the group “hobby”.
[0035]
In addition, information may be classified by adding the number of appearances of all keywords belonging to the group for each group without using weighting and classifying the group into the group having the largest value, or simply selecting the keyword having the highest appearance frequency. They may be classified into groups to which they belong.
[0036]
The melody reproduction process 33 reproduces a sound file according to the group whose information has been classified in the classification process 32. When the group number of the group into which the information is classified is output from the classification process 32, the melody reproduction process 33 acquires the song number set for the group from the group table G, and is specified by the song number in the melody table. Play a sound file.
[0037]
FIG. 7 is a diagram illustrating an example of the melody table M. The melody table M is a table for storing the file name of a sound file, and includes a song number and a file name field for uniquely identifying the sound file.
[0038]
The reproduction timing of the sound file can be set, for example, when receiving information, when browsing information, and the like. Specifically, for example, when an electronic mail arrives, when the received electronic mail is browsed, or when an HTML file is browsed with a Web browser. In the case of reproduction at the time of browsing, it is possible to set so as to change the reproduction volume or the reproduction part of the sound file according to the frequency of appearance of the keyword appearing in the part being displayed. For example, if the keyword appears frequently in the displayed part, the volume is increased or the rust part is set to be reproduced, so that when the information is scrolled and viewed, the volume changes. That is, the change in the appearance frequency of the keyword in the part being displayed can be easily known by the fact that the reproduction part changes to a rust part in the middle.
[0039]
The melody reproduction process 33 may reproduce a sound file set for each keyword in the keyword dictionary W. For example, a sound file set for a keyword having the highest frequency of appearance in the information may be played back. The sound file to be reproduced may be changed according to the keyword that appears most frequently in the section.
[0040]
The volume to be reproduced may be set in advance for each group according to the importance of the group. Also, when an e-mail arrives, if the number of unconfirmed e-mails in the past received e-mails classified into the group into which the e-mails are classified is equal to or more than a predetermined number, the volume to be reproduced is increased and the operator is notified. A warning may be issued.
[0041]
The mapping process 34 displays an asterisk Z at a position on the coordinate plane according to the correlation between the information and the group, as shown in FIG. Specifically, for example, the processing is performed as follows. First, the coordinates for each group are obtained by the following equation.
Σ (set coordinates of group x group evaluation value / total of evaluation values)
Here, the set coordinates of the group are the set coordinates of each group shown in the group table G, and the group evaluation value is a value calculated in the classification process 32. The total of the evaluation values is the total of the group evaluation values of each group.
[0042]
Next, the value obtained by summing the x coordinates for each group is calculated as the x coordinate of the star Z by the above formula, and the value obtained by summing the y coordinates for each group is calculated as the y coordinate of the star Z.
Next, a pop-up window is displayed on the display device 19, and in the pop-up window, an asterisk Z is displayed on an xy coordinate plane whose origin is at the center of the window, and the set coordinates and the origin of each group of preference, hobby, and work are displayed. Display the three connected axes. The timing of the display is, for example, when an electronic mail arrives, or when the transmission of an HTML file is requested to the WWW server 2 and the received HTML file is displayed on a Web browser.
[0043]
For example, it is assumed that the group evaluation value of each group of the received electronic mail is as follows.
Group "Hobbies": 5
Group "work": 3
Group "Taste": 2
In this case, the coordinates calculated for each group by the above equation have the following values.
Group "Hobbies": (160, -100)
Group "work": (-96, -60)
Group "Taste": (0,40)
The sum of these coordinates for each of the x and y coordinates is (64, -120). As a result, a pop-up window displaying the star Z at a position corresponding to (64, -120) on the coordinate plane of the display device 19 is displayed on the display device 19 when an electronic mail arrives.
[0044]
The morphological analysis process 35 performs a process of extracting a word as a component from information output from the communication process 31 and outputting the extracted word to the filter process 37. When the information is output from the communication process 31, the morphological analysis process 35 decomposes the text in the information into words by morphological analysis. Specifically, for example, the text "Slope in western Shizuoka is best for skiing" is broken down into nine words "Shizuoka / western / no / piste / is / ski / ni / optimum /".
[0045]
The kana-kanji conversion process 36 converts the text input by the user based on the kana-kanji conversion dictionary into kana-kanji characters, and outputs the components of the text to be converted to the morphological analysis process 35 for each conversion instruction. Specifically, for example, the operator inputs the text "Slope in the western part of Shizuoka is best for skiing." When the text is divided into five components and conversion is instructed, the kana-kanji conversion process 36 converts the text into a kanji kana mixed sentence of "Shizuoka / Western / Slope / Ski / Best" for each break. , Are output to the morphological analysis process 35. As a result, since relatively small components are input to the morphological analysis process 35, the amount of processing required for decomposing into words in the morphological analysis process 35 can be reduced, and as a result, the process of decomposing information into words can be simplified. Speed can be increased.
[0046]
The kana-kanji conversion process 36 accumulates, for each word, the number of times a word has been converted into a kanji, that is, the frequency of use of a word when creating a text. You may make it output to the morphological analysis process 35.
[0047]
The filter process 37 filters words using the keyword candidate dictionary to exclude words appearing irrespective of the context from the words output from the morphological analysis process 35, and removes a noun whose appearance rate is high only in a specific context from the keyword. Extract as Specifically, for example, three keywords “Shizuoka”, “Slope”, and “Ski” are extracted from nine words “Shizuoka / Seibu / No / Slope / Ha / Ski / Ni / Optimum / is”. .
[0048]
The dictionary update process 38 updates the co-occurrence keyword dictionary V based on the keywords output from the filter process 37 and the keyword dictionary W, and updates the keyword dictionary W based on the co-occurrence keyword dictionary V. FIG. 9 is a diagram illustrating an example of the co-occurrence keyword dictionary V. The co-occurrence keyword dictionary V includes one record including a group number, a registered keyword, a co-occurrence keyword, a co-occurrence count, and an occurrence count. The registered keywords and the co-occurrence keywords in one record are registered in the co-occurrence keyword dictionary V in pairs.
[0049]
The co-occurrence keyword is a keyword that appears (co-occurs) in one piece of information together with a keyword (registered keyword) registered in the keyword dictionary W as a keyword belonging to a group other than “preference” in one piece of information. The co-occurrence keywords are words that are candidates for keywords newly registered in the keyword dictionary W. The extraction conditions of the co-occurrence keyword are, for example, appearing in the information including the registered keyword, appearing in the paragraph including the registered keyword in the information including the registered keyword, and In the included information, it appears in n or less (“n” is an arbitrary value) keywords around the registered keyword. Note that the co-occurrence keywords are not extracted for the keywords belonging to the group “preference”. The number of occurrences is the number of pieces of information including registered keywords that are paired with the co-occurrence keyword among the information to be classified after a certain keyword is stored as a co-occurrence keyword. For the co-occurrence count, the co-occurrence keyword (already stored co-occurrence keyword) is extracted again as the co-occurrence keyword of the registered keyword among the information to be classified after a certain keyword is stored as a co-occurrence keyword. Indicates the number of pieces of information.
[0050]
Note that the number of occurrences may include the number of pieces of information that include the co-occurrence keyword among the information to be classified after the certain keyword is stored as the co-occurrence keyword. Further, the number of occurrences and the number of co-occurrences may be counted based on the number of appearances appearing in the information.
[0051]
The dictionary update process 38 registers a co-occurrence keyword having a strong co-occurrence relationship with the registered keyword as a keyword belonging to a group to which the registered keyword belongs in the keyword dictionary W. The strength of the co-occurrence relationship between the registered keyword and the co-occurrence keyword is, for example, a high probability represented by the number of co-occurrence / occurrence, or a large number of co-occurrence. Specifically, for example, if the co-occurrence keyword is registered in the keyword dictionary W when the conditions of co-occurrence frequency / occurrence frequency> 0.7 and co-occurrence frequency> 9 are satisfied, the co-occurrence keyword dictionary V shown in FIG. If there is a keyword, the newly received e-mail contains the text "Snow is remaining in ski areas in Yamagata Prefecture." And "Snow" is a co-occurrence keyword of the registered keyword "Ski". If it is extracted, “snow” is registered in the keyword dictionary W as the keyword of the group number “1”, and is deleted from the co-occurrence keyword dictionary V. At this time, the set value registered in the keyword dictionary W may be, for example, the same value as the registered keyword paired with the co-occurrence keyword, or represented by the co-occurrence count / occurrence count in the set value of the registered keyword. A value multiplied by the probability may be used. The song number to be registered may be, for example, the same song number as the registered keyword, or may not be set if the sound file set for the group to which the registered keyword belongs should be reproduced. .
[0052]
The dictionary update process 38 accumulates, for each output keyword, the number of pieces of information including the keyword separately from the co-occurrence keyword dictionary V, and when the number satisfies a certain condition, assigns the keyword to the group “preference”. It is added to the keyword dictionary W as a keyword to which it belongs, that is, a keyword whose group number is “0”. When the operator is interested in information including a certain keyword, the operator actively collects information including the keyword, and thus the keyword is likely to be included in a large amount of information generally received by the operator. Therefore, it can be said that the keywords belonging to the group "preferences" are keywords representing the preferences of the operator. By categorizing the keywords into the group "preferences", the operator can easily know that the information is his / her favorite information by the sound at the time of receiving the e-mail.
[0053]
Further, the operator can specify the group of information by himself / herself. Specifically, for example, the processing is performed as follows. When certain information is output from the filter process 37 when the operator is preset to specify a group, the dictionary update process 38 displays an instruction screen for the operator to specify a group into which the information is classified on the display device 19. And the operator specifies a group into which the information is classified. When the operator specifies a group, the dictionary update process 38 additionally registers, in the keyword dictionary W, keywords whose appearance frequency is equal to or more than a predetermined frequency among the extracted keywords as keywords belonging to the group specified by the operator. The operator may be allowed to select a keyword to be additionally registered.
[0054]
The configuration of the information classification device 1 has been described above. Hereinafter, the operation of the information classification device 1 will be described.
FIG. 1 is a flowchart showing a flow of processing of the information classification device 1.
[0055]
First, an operation for initial setting of the melody table M, the group table G, and the keyword dictionary W starting from “Start 4” in the figure will be described. The process of “Start 4” is always performed when the information classification device 1 is used for the first time, and thereafter, is performed as needed. The dictionary update process 38 displays a screen for performing the initial setting of the keyword dictionary W, and requests the operator to input a keyword, a group number to which the keyword belongs, and a set value. When the operator inputs a keyword, a group number to which the keyword belongs, a set value, and a song number, the dictionary update process 38 registers them in the keyword dictionary W (S405). Next, the dictionary update process 38 displays a list of sound file names stored in the melody table M on the screen, and allows the user to select a sound file name for each group. When the operator selects a sound file name, the dictionary update process 38 registers the song number of the selected sound file name in the “song number” field of the group table G (S410). Next, the dictionary update process 38 sets the set coordinates of the group table G according to the number of groups (S415).
[0056]
Next, the operation of updating the dictionary when sending an e-mail starting from “Start 1” in the figure will be described. The operator activates the e-mail program in the information classification device 1, inputs a text to be transmitted, and designates a delimiter by a predetermined keyboard operation, thereby instructing the kana-kanji conversion of the text input so far (S105). When the kana-kanji conversion is instructed, the kana-kanji conversion process 36 converts the instructed text into a kana-kana mixed sentence and outputs the converted component to the morphological analysis process 35 (S110). The morphological analysis process 35 extracts a word and outputs it to the filter process 37. The filter process 37 filters the word to extract a keyword, and outputs the keyword to the dictionary update process 38 (S310). When the keyword is output, the dictionary update process 38 updates the keyword dictionary W and the co-occurrence keyword dictionary V based on the keyword (S315). The above steps S310 to S315 are repeated each time the operator instructs the kana-kanji conversion, and when the operator finishes inputting all the text information and performs an operation of transmitting an e-mail, the communication process 31 converts the e-mail into an e-mail. The data is transmitted to the device 3 (S115).
[0057]
Next, the operation when an e-mail starting from “Start 3” in the figure is received will be described. When receiving the electronic mail from the electronic mail device 3 via the communication unit 15, the communication process 31 outputs the text information of the received electronic mail to the morphological analysis process 35 (S305). The morphological analysis process 35 decomposes the text information into words and outputs the words to the filter process 37. The filter process 37 outputs the keyword extracted from the word to the dictionary update process 38 (S310). When a keyword is input, the dictionary update process 38 updates the keyword dictionary W and the co-occurrence keyword dictionary V based on the keyword (S315). Next, the classification process 32 classifies the text information of the e-mail output from the communication process 31 into groups, and outputs the group to the melody reproduction process 33 and the mapping process 34 (S320). The melody reproduction process 33 performs a process of reproducing sounds according to the classified groups. For example, if the classified group is “hobby”, the melody reproduction process 33 acquires the file name “song name 1.mmf” corresponding to the classified group “hobby” from the group table G, and stores the sound file in the external storage device. 20. Next, the acquired sound file is output to the sound control unit 16. As a result, the sound file is reproduced by the sound control unit 16, and the sound corresponding to the group into which the electronic mail is classified is reproduced. On the other hand, the mapping process 34 displays a pop-up window in which a star is displayed at a position corresponding to the correlation between the information and the group on the display device 19 (S325). Next, the e-mail program displays the e-mail on the display device 19. The melody reproduction process 33 reproduces the sound file and controls the reproduction volume in accordance with the appearance frequency of the registered keyword included in the part of the text information being displayed.
[0058]
The operation when browsing the Web starting from “Start 2” in the figure, that is, when receiving the HTML file, is substantially the same as the operation when “Start 3”, so the description is omitted.
[0059]
Hereinafter, effects of the present embodiment will be described.
According to the information classification device 1 of the present embodiment, a co-occurrence keyword having a strong co-occurrence relationship with a registered keyword is newly registered in the keyword dictionary W. A strong co-occurrence relationship means that the co-occurrence keyword frequently appears in the information where the registered keyword appears. When the co-occurrence keyword appears in certain information, the information is It is highly likely that the information belongs to the group. Further, even if the target of interest shifts with the passage of time, it is highly likely that the user will be interested in information related to information that has been interested in the past in the future. When a co-occurrence keyword is registered as a keyword in the keyword dictionary W, information in which the original registered keyword does not appear but only the newly registered keyword appears can be classified into a group to which the original registered keyword belongs. Therefore, information can be correctly classified into groups.
[0060]
Further, information is classified into groups having a large group evaluation value calculated based on weighting for each keyword, so that information can be more accurately classified into groups. Further, when the operator is not set to designate a group, there is no need for the operator himself to register keywords and classify information, so that the information can be easily classified into groups. For example, even if the importance of information changes over time, if the frequency of occurrence of a specific word increases with time, the information classification device 1 registers the word as a keyword, and the operator himself , The information having high importance can be classified into a specific group without continuously updating the keyword dictionary W. Further, since the melody reproduction process 33 reproduces a sound according to the classified group, the group of information can be notified to the operator by sound. Thus, the operator can judge the group of information based on the sound, and, for example, in the case of an e-mail, the operator can immediately know the arrival of an important e-mail.
[0061]
Further, the information classification device 1 according to the present embodiment extracts the constituent elements of the information for each section of the text input when the information is created by the text input, so that the kana-kanji conversion process 36 The processing in the morphological analysis process 35 is simplified for the components of
[0062]
Further, the information classification device 1 of the present embodiment displays a position corresponding to the correlation between the information and the group on a coordinate plane. Accordingly, when one information includes a plurality of keywords and the groups to which the keywords belong are different, the degree of the relationship between the information and the plurality of groups is visually expressed. Therefore, the tendency of the information correlated with the plurality of groups can be accurately transmitted to the operator.
[0063]
Furthermore, according to the information classification device 1 of the present embodiment, the information is classified into groups according to the instruction of the operator, so that the intention of the operator can be directly reflected in the classification processing.
[0064]
In the present embodiment, the e-mail and the HTML file are described as examples of the information to be classified. However, the information may be an image file or a sound file. For example, the sound files are grouped based on the file name of the sound file. It may be classified.
[Brief description of the drawings]
FIG. 1 is a flowchart illustrating a flow of processing of an information classification device according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating a state where an information classification device according to an embodiment of the present invention is connected to a network.
FIG. 3 is a block diagram illustrating a hardware configuration of an information classification device according to an embodiment of the present invention.
FIG. 4 is a data flow diagram of the information classification device according to one embodiment of the present invention.
FIG. 5 is a diagram showing a keyword dictionary included in the information classification device according to one embodiment of the present invention.
FIG. 6 is a diagram showing a group table provided in the information classification device according to one embodiment of the present invention.
FIG. 7 is a diagram showing a melody table provided in the information classification device according to one embodiment of the present invention.
FIG. 8 is an example in which an information classification device according to an embodiment of the present invention displays a position corresponding to a correlation between information and a group on a coordinate plane.
FIG. 9 is a diagram showing a co-occurrence keyword dictionary included in the information classification device according to one embodiment of the present invention.
[Explanation of symbols]
1 Information classification device
2 WWW server
3 E-mail device
11 CPU (sound reproducing means, dictionary updating means, manual classifying means, weighting means, classifying means, mapping means)
12 ROM
13 RAM (sound reproducing means, dictionary updating means, manual classifying means, weighting means, classifying means, mapping means)
14 Actuator (manual classification means)
15 Communication unit
16. Sound control unit (sound reproducing means)
17 speakers (sound reproducing means)
18 Display control unit (manual classification means, weighting means, mapping means)
19 Display device (manual classification means, weighting means, mapping means)
20 external storage device (sound reproducing means, dictionary updating means, weighting means)
31 Communication Process
32 Classification Process
33 Melody Playback Process
34 Mapping Process
35 Morphological Analysis Process
36 Kana-Kanji conversion process
37 Filter Process
38 Dictionary update process

Claims (11)

Classification means for classifying information into groups based on the keyword dictionary;
Dictionary updating means for updating the keyword dictionary based on components extracted from information;
Sound reproducing means for reproducing sound according to a group into which information is classified;
An information classification device, comprising: 2. The information classification apparatus according to claim 1, wherein the dictionary updating unit extracts a component of the information for each section of the text input when the information is created by the text input. 3. The information classification device according to claim 1, wherein the dictionary updating unit extracts a component of the information from the received information. The dictionary updating means extracts from the information a component having a strong co-occurrence relationship with a keyword registered in the keyword dictionary and registers the component in the keyword dictionary as a paired keyword and a paired keyword. The information classification device according to claim 1, 2 or 3. The information classification device according to any one of claims 1 to 4, wherein, when the information is received, the sound reproduction unit reproduces a sound according to a group into which the information is classified. Weighting means for registering weighting for the keyword in the keyword dictionary,
The information classification device according to any one of claims 1 to 5, wherein the classification unit classifies information into groups based on weights registered in keywords. The sound reproducing unit changes control information for reproducing a sound in accordance with an appearance frequency of a keyword registered in the keyword dictionary and an appearance frequency of the information in a displayed portion. Item 7. The information classification device according to any one of Items 1 to 6. The information classification device according to any one of claims 1 to 7, further comprising mapping means for displaying a position corresponding to the correlation between the information and the group on a coordinate plane. The information according to any one of claims 1 to 8, wherein the dictionary update unit additionally registers a component extracted from the information as a keyword paired with a group specified by an operator in the keyword dictionary. Classifier. A classification step of classifying information into groups based on the keyword dictionary;
A dictionary update step of updating the keyword dictionary based on the components extracted from the information,
A sound reproduction step of reproducing a sound according to a group into which information is classified;
An information classification method comprising: A program for causing a computer to function as an information classification device,
Classification means for classifying information into groups based on the keyword dictionary,
Dictionary updating means for updating the keyword dictionary based on components extracted from information;
Sound reproducing means for reproducing sound according to a group into which information is classified;
Program to function as

Images