JP6260208B2 - Text summarization device - Google Patents

Description

  The present invention relates to a text summarization apparatus that summarizes input text data and generates summary text data.

  An apparatus that summarizes input text data and reads it out as speech is known (for example, see Patent Document 1). In such a summary reading device, summarization is performed so that the summarization rate set by the user is obtained using the importance given to each word included in the input text.

  Also, the input text data is divided into a plurality of partial texts and a plurality of importance levels corresponding to the partial texts, and the content of the less important partial texts is skipped based on the set speed command for speed reading. A speech synthesizer is known (see, for example, Patent Document 2).

JP 2001-282815 A JP-A-5-181491

  However, in the conventional apparatus, although the degree of summarization in the input text can be changed by changing the summarization rate or speed command, the importance given to the partial text constituting the text data is dynamically changed according to the situation. Could not change. Therefore, even partial text that is set to low importance may not be included in the summary text, although it should be provided to the user in some situations. Even if the partial text is set to a high degree, it may not be provided to the user depending on the situation, but may be provided in the summary text. There was a problem that information could not be provided appropriately.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a text summarization apparatus that can appropriately provide summary information to a user.

The text summarization device according to the present invention stores a data input unit to which text data composed of a plurality of partial text data is input, and an importance level for storing the importance level assigned to the partial text data input to the data input unit. From the storage unit, the importance level changing unit that changes the importance level based on the history information of the partial text data included in the text data input in the past, and the text data input to the data input unit based on the importance level A data processing unit that extracts one or a plurality of partial text data to generate summary text data, and the importance level changing unit inputs in the past when the text data input in the past is text data related to emergency information The importance corresponding to the partial text data included in the text data is increased.

  According to the text summarizing apparatus of the present invention, since the importance level is changed based on the history information of the partial text data included in the past text data, it is possible to appropriately provide the summary information to the user.

It is a figure which shows the structural example of the text summarization apparatus which concerns on Embodiment 1. FIG. The flowchart which shows the operation example about the update of the importance which concerns on Embodiment 1 It is a figure which shows the example of the input text and summary text which concern on Embodiment 1. FIG. 6 is a diagram illustrating an example of newly input text according to Embodiment 1. FIG. It is a figure which shows the other structural example of the text summarizing apparatus which concerns on Embodiment 1. FIG. It is a figure which shows the other structural example of the text summarizing apparatus which concerns on Embodiment 1. FIG. It is a figure which shows the other structural example of the text summarizing apparatus which concerns on Embodiment 1. FIG. It is a figure which shows the structural example of the text summarization apparatus which concerns on Embodiment 2. FIG. It is a figure which shows the example of the input text 101 which concerns on Embodiment 2. FIG. 10 is a flowchart illustrating an operation example of an analysis unit 21 according to the second embodiment. It is a figure which shows the example of the analysis result text 102 which concerns on Embodiment 2. FIG. 10 is a flowchart illustrating an operation example of the importance level assigning unit 22 according to the second embodiment. It is a figure which shows the example of the importance level table 103 which concerns on Embodiment 2. FIG. It is a figure which shows the example of the text 104 with the importance which concerns on Embodiment 2. FIG. 10 is a flowchart illustrating an operation example of a partial text data selection unit 23 according to the second embodiment. It is a figure which shows the example of the summary text 105 which concerns on Embodiment 2. FIG. It is a figure which shows the example of the text log | history 106 which concerns on Embodiment 2. FIG. It is a figure which shows the other structural example of the text summarizing apparatus based on Embodiment 2. FIG. It is a figure which shows the other structural example of the text summarizing apparatus based on Embodiment 2. FIG. It is a figure which shows the other structural example of the text summarizing apparatus based on Embodiment 2. FIG.

Embodiment 1 FIG.
Embodiment 1 of the present invention will be described below with reference to the drawings.

  FIG. 1 is a diagram illustrating a configuration example of a text summarizing apparatus according to the first embodiment. The text summarizing apparatus 100 includes a text data input unit 1, a data processing unit 2, an importance level storage unit 3, a summary level change unit 4, a text history data storage unit 5, and an importance level change unit 6. The text summarization device 100 corresponds to, for example, a device mounted on the navigation device or the navigation device itself, but is not limited to this. The text data is input and summary information that is the content of the summary text is input. Anything provided to the user is acceptable. The provision of the summary information is not limited to providing the contents of the summary text as a document, but also includes providing the contents of the summary text as speech.

  Text data is input to the text data input unit 1. The text data input here corresponds to, for example, document data representing contents such as Web information such as news, emergency information such as earthquake early warning, weather information, and peripheral facility information. The text data is composed of a plurality of partial text data. The partial text corresponds to, for example, a sentence, a phrase constituting the sentence, or a word constituting the phrase. The text data is input to the text data input unit 1 when the text summarizing apparatus 100 requests the server or the like, for example.

  The data processing unit 2 extracts one or more partial text data from the text data input to the data input unit 1 based on the importance given to the partial text data constituting the input text data, and outputs the summary text data. Generate. Here, the importance is an index representing the degree to be presented to the user. Therefore, it can be said that partial text data with high importance among text data is data to be presented to the user by being included in the summary text data.

  The importance storage unit 3 stores partial text data and importance in association with each other, and is realized by, for example, a memory. The importance may be set, for example, by learning from the number of appearances of words in a large number of texts input in the past, or may be arbitrarily set by the user. Even if the partial text is a sentence or clause, the sum of the importance of the words that make up the sentence or clause, or the normalization by dividing the sum of importance by the number of words, etc. Therefore, it can be said that the importance storage unit 3 stores the importance corresponding to the partial text data as a result. The information on the importance stored in the importance storage unit 3 may be configured to be held by the data processing unit 2, for example. When the partial text is a word, the importance may be stored in consideration of co-occurrence with the word as well as the word included in the text input in the past.

  Further, the importance may be set from the number of appearances of a word in one input text. Further, a value obtained by TF-IDF (Term Frequency-Inverse Document Frequency) may be used as the importance. Further, the importance level storage unit 3 may store the part of speech information together with the word information, and may increase the importance level of nouns and adjectives that characterize the document.

  The summarization level changing unit 4 changes the value of the set summarization level. The summarization degree is an index representing the degree of summarization presented to the user. The higher the value, the shorter the summary text provided to the user. For example, if the text summarization device is a navigation device, the user can set the degree of summarization using a dial or a button. However, the summarization level does not necessarily have to be changed, and the data processing unit 2 may store a preset summarization level as a fixed value. In the following description, the summarization level is described as a fixed value.

  The text history data storage unit 5 stores the partial text data extracted by the data processing unit 2 and the number of appearances thereof as text history data (history information). In the following description, partial text data and the number of appearances thereof are described as history information, but the present invention is not limited to this. For example, a weighting value or the like for changing the importance calculated based on the number of appearances may be stored in the text history data storage unit 5 as history information. When the partial text data is a sentence or a phrase, the number of appearances for each word constituting the sentence or the phrase may be stored in the text history data storage unit 5 as history information.

  The importance level changing unit 6 stores the importance level of one or more partial text data extracted by the data processing unit 2 out of the importance levels of the partial text data stored in the importance level storage unit 3. It changes based on the history information of the partial text data stored in the part 5.

  Next, the operation of the importance level changing process in the first embodiment will be described. FIG. 2 is a flowchart showing an operation example for changing the importance according to the first embodiment.

  First, text data is input to the text data input unit 1 (step ST1). FIG. 3 is a diagram showing an example of input text and summary text according to the first embodiment. As shown in FIG. 3, the input text is composed of a plurality of partial texts. The partial text 1 is “ABCDEFG”, the partial text 2 is “HIJKLMN”, and the partial text 3 is “OPQRSTU”.

  Next, the data processing unit 2 extracts partial text data from the input text data (step ST2), acquires the importance of the extracted partial text data from the importance storage unit 3, and compares it with the summarization ( Step ST3). In the example of FIG. 3, the data processing unit 2 extracts partial text data corresponding to the partial text 1 from the input text data, and acquires the corresponding importance from the importance storage unit 3. Here, the importance of the partial text data corresponding to the partial text 1 is 3.5, and the summarization is 3.0. Hereinafter, the importance assigned to the partial text data may be simply expressed as the importance of the partial text.

  When the importance of the extracted partial text data is higher than the summarization degree (step ST4-Yes), the data processing unit 2 stores the partial text data and the number of appearances thereof as history information in the text history data storage unit 5 ( Step ST5). In the example of FIG. 3, since the importance of the partial text 1 is higher than the summarization degree, the data processing unit 2 stores the partial text 1 “ABCDEFG” and the number of appearances “1” in the text history data storage unit 5 as history information. .

  When there is remaining partial text data (step ST6-Yes), the data processing unit 2 performs the processing from step ST2 to step ST5 again. In the example of FIG. 3, since the remaining partial text data other than the partial text 1 remains, the data processing unit 2 performs the same process as the partial text 1 on the data of the next partial text 2. Next, the same processing as that for the partial texts 1 and 2 is performed on the data of the partial text 3. Here, the importance of the partial text 2 is 2.0, which is lower than the summary degree 3.0, and the importance of the partial text 3 is 3.2, which is higher than the summary degree 3. Therefore, the data processing unit 2 extracts the data of the partial text 1 and the partial text 3 from the input text data. As a result, the text history data storage unit 5 stores the partial text 1 “ABCDEFG”, the appearance count “1”, and the partial text. 3 “OPQRSTU” and appearance count “1” are stored as history information.

  If there is no remaining partial text data (step ST6-No), the data processing unit 2 creates summary text data (step ST7). In the example of FIG. 3, since the partial text data of the partial texts 1 and 3 are extracted from the input text data, the summary text data corresponding to the summary text “ABCDEFG OPQRSTU” is created.

  The importance level changing unit 6 changes the importance level stored in the importance level storage unit 3 based on the history information of the partial text data stored in the text history data storage unit 5 (step ST8). In the example of FIG. 3, the text history data storage unit 5 stores the number of appearances “1” of the partial text 1 “ABCDEFG” and the partial text 3 “OPQRSTU”, and the importance level changing unit 6 stores the importance level memory. Assume that the importance levels of partial texts “ABCDEFG” and “OPQRSTU” stored in part 3 are changed to 2.5 and 2.2, respectively.

  The importance level changing unit 6 changes the importance level of the partial text data included in the text data if the input text data is important information for the user, and changes the level of the partial text data if it is not important for the user. Change to be less important. Although details will be described later, here, the importance level changing unit 6 increases the importance of the partial text data stored in the importance level storage unit 3 as the number of appearances of the partial text data stored in the text history data storage unit 5 increases. It demonstrates as what changes so that a degree may become low.

  Here, the change in the importance level when the text data shown in FIG. 3 is input has been described, but this importance level is updated every time text data is newly input and summary text data is created. . For example, when new text data including partial text “ABCDEFG” is input and extracted as summary text data, the number of occurrences of “ABCDEFG” stored in the text history data storage unit 5 is changed to “2”. Then, the importance level changing unit 6 changes the importance level of the partial text “ABCDEFG” stored in the importance level storage unit 3 based on the appearance count “2” so that the importance level is further lowered.

  In FIG. 2, it has been described that the importance level is changed in step ST8 after the summary text data is created in step ST7. However, the order of these processes is arbitrary, or these processes may be performed in parallel. . In steps ST3 to ST7, the importance level and the summarization level of the extracted partial text data are compared. When the importance level is higher than the summarization level, history information is stored in the text history data storage unit 5, and the summary text data is stored. However, the present invention is not limited to this. For example, even if the importance level and the summary level are not compared, n (n is an arbitrary integer) pieces of relatively high importance partial text data are extracted from the partial text data constituting the input text data. Thus, summary text data may be created and history information may be stored in the text history data storage unit 5. In that case, the summarization level is not necessary in the extraction of the partial text data.

  Next, the operation of the summary text data generation process when new text data is input will be described. FIG. 4 is a diagram illustrating an example of newly input text according to the first embodiment. The text data shown in FIG. 3 is called text data input in the past in the following description in order to distinguish it from the newly input text data shown in FIG. The input text of FIG. 3 and the input text of FIG. 4 have similar contents, but the input text of FIG. 4 includes a partial text “VWXYZ” that is different from the text previously input shown in FIG. Is different. In addition, as a situation where text data having similar contents is input, for example, a situation in which Web news or the like having similar contents but a different source is input is assumed.

  When new text data is input, first, processing from steps ST1 to ST3 in FIG. 2 is performed. In step ST3, the data processing unit 2 compares the importance level of the partial text data extracted from the newly input text data with the summary level. The importance here is a value changed by the importance changing unit 6 based on the history information of the partial text data included in the text data input in the past. In the example of FIG. 4, the importance levels of the partial text 1 and the partial text 3 are changed to low values of 2.5 and 2.2 based on the history information of the partial text 1 and the partial text 3 input in the past, respectively. The sum is less than 3.0. Further, since the importance of the partial text 2 is not changed by the text data input in the past, it remains 2.0. The importance of the partial text 4 “VWXYZ” is 2.8.

  Since the importance level of the partial text data constituting the newly input text data is not higher than the summarization level, the processes after step ST4 are not performed. Therefore, the data processing unit 2 does not generate the summary text data from the newly input text data, and therefore repeatedly provides the summary text having the same or similar contents as the summary text generated from the previously input text data. Therefore, summary information can be appropriately provided to the user.

  Heretofore, the example in which the importance level changing unit 6 reduces the importance level of the partial text based on the history information of the text data input in the past has been described. However, the importance level changing unit 6 may be changed to increase the importance level. Good. By doing so, the importance of the partial text data constituting the newly input text data is changed to a high level even when text data similar in content to the text data input in the past is newly input. Therefore, the data processing unit 2 can generate a summary text having the same or similar content as the past summary text and provide it to the user.

  In particular, when the input text data is text data related to emergency information such as an earthquake early warning, it is often necessary to repeatedly provide the text data even if it has been provided to the user in the past. Even in such a case, the contents of the summary text relating to the emergency information can be repeatedly provided, and the summary information can be appropriately provided to the user.

  As described above, according to the first embodiment, the importance level changing unit 6 stores the partial text data stored in the importance level storage unit 3 based on the history information of the partial text data included in the text data input in the past. Since the importance level of the text is changed, when new text data is input, the summary information of the text is actively provided if necessary and not provided unless necessary. It is possible to provide summary information appropriately.

  FIG. 5 is a diagram showing another configuration example of the text summarizing apparatus according to the first embodiment. As shown in FIG. 5, the text summarizing device 110 may include a speech synthesis unit (speech generation unit) 7.

  The speech synthesizer 7 synthesizes the content of the summary text based on the summary text data generated by the data processing unit 2 and outputs it to the outside. In the example of FIG. 3, “ABCDEFG OPQRSTU” is provided to the user as voice.

  By doing so, since the user can listen to the summary content by voice, for example, if the text summarization device is a navigation device, the user can be provided with summary information without looking at the navigation screen during driving, Safe driving is possible.

  FIG. 6 is a diagram showing another configuration example of the text summarizing apparatus according to the first embodiment. As shown in FIG. 6, the text summarizing device 120 may include an operation history storage unit 8 and a preference keyword extraction unit 9.

  The operation history storage unit 8 stores past user operation history. For example, in the navigation device, the user operation history corresponds to a destination setting operation, selection of a music CD to be played in the vehicle, and the like.

  The preference keyword extraction unit 9 extracts a keyword representing the user's preference from the operation history information stored in the operation history storage unit 8. For example, when the user selects and operates a song of artist “XXX” as a music CD, the preference keyword extraction unit 9 extracts “XXX” as a preference keyword and stores it as history information in the text history data storage unit 5. Based on the history information stored in the text history data storage unit 5, the importance level changing unit 6 configures text data corresponding to the preference keyword among the importance levels of the partial text data stored in the importance level storage unit 3. Change the partial text data so that the importance is high.

  By doing so, when text data related to the artist “XXX” is input, the summary text data is easily generated, and it is possible to appropriately provide the summary information about the items that the user is interested in.

  FIG. 7 is a diagram showing another configuration example of the text summarizing apparatus according to the first embodiment. As shown in FIG. 7, the text summarizing device 130 may include a speech recognition keyword extracting unit 10.

  The speech recognition keyword extraction unit 10 recognizes an external speech, extracts speech recognition information as text data, and stores history information of partial text data constituting the text data in the text history data storage unit 5. Here, the voice from the outside corresponds to, for example, a conversation inside or outside the vehicle, the contents of the radio, or a voice by CD. Such external audio can be considered as content of interest by the user. Therefore, the importance level changing unit 6 extracts the importance level of the partial text data stored in the importance level storage unit 3 by the voice recognition keyword extraction unit 10 based on the history information stored in the text history data storage unit 5. It changes so that the importance of the partial text data which comprises the text data corresponding to the made keyword becomes high.

  By doing so, when text data related to a speech-recognized keyword is input, the summary text data can be easily generated, and it is possible to appropriately provide summary information about a matter of high user interest. . Note that when the speech-recognized keyword is considered as a topic that has already been discussed, the importance level changing unit 6 may change the keyword so that the importance level of the partial text data related to the keyword is lowered.

Embodiment 2. FIG.
The second embodiment of the present invention will be described below with reference to the drawings.

  FIG. 8 is a diagram showing a configuration example of the text summarizing apparatus according to the second embodiment. The text summarization apparatus 200 according to the second embodiment is implemented in that the data processing unit 2 includes an analysis unit 21, an importance level assigning unit 22, a summary text data selection unit 23, and a summary text data storage unit 24. Different from the text summarization apparatus 100 of the first embodiment. Since the other configuration is the same as that of the first embodiment, the same reference numerals as those in FIG.

  The analysis unit 21 performs sentence analysis (language analysis) of the text data input to the text data input unit 1. That is, the analysis unit 21 divides the text data into partial text data.

  The importance level assigning unit 22 applies the importance level stored in the importance level storage unit 3 to the partial text data generated by the analysis unit 21 with respect to the partial text data analyzed by the analysis unit 21. Give importance.

  The partial text data selection unit 23 selects partial text data having an importance level higher than the summary level input from the summary level change unit 4 among the partial text data levels assigned with the importance level by the importance level assignment unit 22. Select to include in summary text data.

  The summary text data storage unit 24 stores summary text data composed of partial text data selected by the partial text data selection unit 23.

  Next, the operation from the input of text data until the history information of partial text data is stored in the second embodiment will be described. FIG. 9 is a diagram illustrating an example of the input text 101. In the following description, the partial text is described as being expressed in phrase units. However, as described in the first embodiment, the partial text may be in sentence units or in word units.

  FIG. 10 is a flowchart showing an operation example of the analysis unit 21 according to the second embodiment. First, the analysis unit 21 divides the text shown in FIG. 9 input to the text data input unit 1 into sentences (step ST21). The division into sentences can be realized, for example, by dividing at a punctuation point.

  Next, the analysis unit 21 further divides the text divided into sentences into segment units (step ST22). For example, a syntax analyzer such as KNP or CaboCha may be used to divide sentences into phrases. The syntax analyzer analyzes what structure a sentence is composed of and outputs the structure.

  Subsequently, the analysis unit 21 further divides the text divided into phrases into word units (step ST23). For division into phrases from words, for example, a morphological analyzer such as Mecab may be used. Note that the morphological analyzer analyzes what words and parts of speech a sentence consists of.

  The analysis unit 21 creates an analysis result text as a result of language analysis. FIG. 11 is a diagram illustrating an example of the analysis result text 102 according to the second embodiment. As shown in FIG. 11, the input text 101 is divided into word units as “new model / rocket /“ / epsilon / ”...” As a result of analysis by the analysis unit 21. In FIG. 11, “/” represents a word break, “//” represents a sentence break, and “///” represents a sentence break.

  FIG. 12 is a flowchart showing an operation example of the importance level assigning unit 22 according to the second embodiment. First, analyzed text data divided into word units by the analysis unit 21 is input to the importance level assigning unit 22 (step ST31).

  Next, the importance assigning unit 22 assigns importance to each word divided by the analysis unit 21 using the importance of the importance table stored in the importance storage unit 3 (step ST32). FIG. 13 is a diagram showing an example of the importance level table 103. In the example of FIG. 13, importance is stored in the importance storage unit 3 in correspondence with words. For example, the importance level 15 is given for the word “new type”, and the importance level 2 is given for the word “ramen”.

  The importance level assigning unit 22 creates a text with importance level by giving importance to the analysis result text 102. FIG. 14 is a diagram showing an example of the text 104 with importance according to the second embodiment. As shown in FIG. 14, the importance of the first phrase “New rocket“ Epsilon ”is the first” is the words “New”, “Rocket”, ““ ”,“ Epsilon ”,“ ””, “First” The sum of the importance given to each of “No.”, “Machine”, and “Ga” is divided by the number of words and normalized to 0.7. Similarly, the importance of the second phrase “2pm on the 14th” is 0.2, and the third phrase “at the Uchinoura Space Observatory, Japan Aerospace Exploration Agency in Kappacho, Kagoshima” The importance is “0.4”, and the fourth phrase “launched” has an importance of 1.0. As described above, the importance level assigning unit 22 gives the importance level to the clause as the partial text.

  In addition, the importance level assigning unit 22 normalizes the sum of the importance levels of each clause by dividing the number of clauses, so that the first sentence of the new rocket “Epsilon” is 2:00 pm The importance of “launched at the Ueno Space Research Station, Japan Aerospace Exploration Agency” can be calculated as 0.6. Although the importance is obtained here by normalization, the present invention is not limited to this.

  FIG. 15 is a flowchart showing an operation example of the partial text data selection unit 23 according to the second embodiment. The partial text data selection unit 23 first compares the sentence summarization level input from the summarization level changing unit 4 with the sentence importance level (step ST41). As a result of the comparison, if the importance of the sentence is equal to or higher than the sentence summarization (step ST41-Yes), the text data selection unit 23 performs the process of step ST42, while the sentence importance is higher than the sentence summarization degree. Is smaller (step ST41-No), the process proceeds to step ST46. Here, the sentence summarization level is 0.5. Then, in the example of FIG. 14, since the importance level of the first sentence is 0.6, which is equal to or higher than the sentence summarization level, the process proceeds to step ST42. Note that the summarization level may be a value held in advance by the partial text data selection unit 23 instead of the value input from the summarization level changing unit 4.

  Next, the partial text data selection unit 23 compares the importance level of the phrase included in the sentence selected in step ST41 with the summary level of the phrase (step ST42). The phrase summarization level may be the same value as the sentence summarization level, or may be set differently from the sentence summarization level. Here, the summarization level of the clause is 0.5, which is the same as the summarization level of the sentence. In the example of FIG. 14, the importance of the first phrase “new rocket“ Epsilon ”first aircraft” is 0.6, which is higher than the summarization level (step ST43—Yes). The corresponding partial text data is stored in the summary text data storage unit 24 (step ST43), and the partial text data corresponding to the content of the first phrase and the number of appearances thereof are stored in the text history data storage unit 5 as history information. (Step ST44).

  If the sentence to be compared between the importance level and the summary level by the partial text data selection unit 23 is the last, the process ends (step ST46—Yes). Here, since it is not the last sentence (step ST46-No), a process transfers to the following clause (step ST45).

  Next, the partial text data selection unit 23 compares the importance level of the second phrase “2pm on the 14th” with the summary level of the phrase (step ST42). Since the importance level of the second clause is 0.2, which is smaller than the summary level of the clause (step ST42-No), the processing moves to the next clause (step ST45). The third phrase “At Uchinoura Space Observatory of the Japan Aerospace Exploration Agency in Katsukijima, Kagoshima Prefecture” also has an importance of 0.4, which is smaller than the summarization of the phrase, so the same processing as the second phrase Is made. Since the importance level of the fourth phrase “Launched” is 1.0, which is equal to or higher than the summary level of the phrase, the same processing as the first phrase is performed.

  For the second sentence and after, the same process as the first sentence is performed, and if the same process is performed for all sentences in the input text (step ST46-Yes), the process of the partial text data selection unit 23 is performed. Ends.

  As a result of the partial text selection process by the partial text data selection unit 23, the summary text data is stored in the summary text data storage unit 24. FIG. 16 is a diagram illustrating an example of the summary text 105 according to the second embodiment. As shown in FIG. 16, the summary text is the partial text selected by the partial text data selection unit 23, that is, the first clause “new rocket“ Epsilon ”first machine” and the fourth clause “launched”. It is composed of.

  Further, the history information is recorded by the partial text data selection unit 23 (step ST44), and the text history data storage unit 5 stores the number of appearances of the partial text data as the text history. FIG. 17 is a diagram showing an example of the text history 106 according to the second embodiment. As shown in FIG. 17, the number of appearances “1” for the words “new”, “rocket”, “epsilon”, “first issue”, and “machine” in the first phrase is stored. In addition, the number of appearances “1” for the word “launch” in the fourth phrase is stored. The same applies to the second and subsequent sentences.

  As described above, according to the second embodiment, the importance level is given by the importance level assigning unit 22 to the partial text data divided by the language analysis by the analysis unit 21, and the partial text data is based on the importance level. Since the selection unit 23 selects the partial text data and stores the history information in the history data storage unit 5, the same effects as those of the first embodiment are obtained. Moreover, since the analysis unit 21 performs language analysis and divides the input text data into partial text data, it is possible to store history information of the partial text data in which the sentence structure is considered.

  FIG. 18 is a diagram showing another configuration example of the text summarizing apparatus according to the second embodiment. As shown in FIG. 18, the text summarization device 210 according to the second embodiment may also include the speech synthesizer 7. Note that since the speech synthesizer 7 of the second embodiment is the same as that of the first embodiment, the same reference numerals as those in FIG.

  FIG. 19 is a diagram showing another configuration example of the text summarizing apparatus according to the second embodiment. As shown in FIG. 19, the text summarization device 220 of the second embodiment may also include a preference keyword extraction unit 9. Note that the preference keyword extraction unit 9 of the second embodiment is the same as that of the first embodiment, so the same reference numerals as those in FIG.

  FIG. 20 is a diagram showing another configuration example of the text summarizing apparatus according to the second embodiment. As shown in FIG. 20, the text summarization device 230 of the second embodiment may also include the speech recognition keyword extraction unit 10. Note that the speech recognition keyword extraction unit 10 of the second embodiment is the same as that of the first embodiment, and therefore the same reference numerals as those in FIG.

  DESCRIPTION OF SYMBOLS 1 Text data input part, 2 Data processing part, 3 Importance storage part, 4 Summarization degree change part, 5 Text history data storage part, 6 Importance change part, 7 Speech synthesizer, 8 Operation history storage part, 9 Preference keyword Extraction unit, 10 speech recognition keyword extraction unit, 21 analysis unit, 22 importance assigning unit, 23 partial text data selection unit, 24 summary text data storage unit, 100, 110, 120, 130, 200, 210, 220, 230 text Summary device

Claims (7)

A data input part for inputting text data composed of a plurality of partial text data; and
An importance storage unit for storing importance assigned to the partial text data input to the data input unit;
An importance changing unit that changes the importance based on history information of the partial text data included in the text data input in the past;
A data processing unit that extracts one or more partial text data from text data input to the data input unit based on the importance and generates summary text data;
The importance level changing unit increases the importance level corresponding to partial text data included in the text data input in the past when the text data input in the past is text data related to emergency information. A text summarization device. The importance level changing unit increases the importance level of the partial text data included in the text data input in the past when the information of the text data input in the past is information important to the user, to claim 1, characterized in that when the information of the text data entered into the past is not important information for the user to lower the importance of the partial text data contained in the text data inputted to the past Description text summarization device. When the text data input in the past is text data related to Web information, the importance level changing unit adds the text data input in the past out of the importance levels stored in the importance level storage unit. The text summarization apparatus according to claim 2 , wherein the importance corresponding to the partial text data included is lowered. A preference information extraction unit that extracts user preference information as text data from past operation history by the user,
The importance level changing unit corresponds to the importance level corresponding to the partial text data included in the text data related to the preference information extracted by the preference information extraction unit among the importance levels stored in the importance level storage unit. The text summarization apparatus according to claim 2, wherein the text summarization apparatus is made high. A speech recognition information extraction unit that recognizes external speech and extracts speech recognition information as text data;
The importance level changing unit corresponds to the partial text data included in text data related to the speech recognition information extracted by the speech recognition information extraction unit among the importance levels stored in the importance level storage unit. 5. The text summarization apparatus according to claim 2, wherein the importance is increased. The data processing unit
An analysis unit that analyzes the text data input to the data input unit and divides the text data into a plurality of partial text data that are sentences, clauses, or words;
An importance level assigning unit that gives the importance level stored in the importance level storage unit to the plurality of partial text data divided by the analysis unit;
A partial text data selection unit that selects the partial text data having a value that is greater than a set value by the importance level assigned by the importance level grant unit from among the plurality of partial text data. The text summarization device according to any one of claims 2 to 5 . Text summarizing apparatus according to any one of claims 1 to 6, characterized in that it comprises a speech synthesis unit to output the speech synthesis the contents of the short text based on the summary text data generated by the data processing unit.

Images