JP2766261B2 - Sentence summarization system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention] (Industrial application field) The present invention relates to a text summarization system for summarizing text information using a computer. (Prior art) In recent years, the technology of natural language understanding by a computer has been improved, and a machine translation system that translates a sentence into another language one sentence at a time, and a summary that grasps the entire flow of a sentence and outputs only the essential points from the whole. The system is being developed. This summarization system models the contents of a sentence, extracts important parts from the model, expresses the result in natural language, and outputs the result. However, sentences generally handled by humans have various formats,
Content is mixed. For example, if you consider a general paper, the noun phrase of the concise expression appears in the title part,
In the text, there is a part that outlines and a part that describes individual details. For this reason, it is not meaningful to uniformly summarize one sentence in which various forms and contents are mixed. (Problems to be Solved by the Invention) As described above, the conventional summarization system has a drawback that only a uniform summarization result can be obtained because the summarization processing method or the text summarization processing range is fixed. there were. Therefore, an object of the present invention is to change the summarization method for each part appearing in one sentence according to its format and content, and to output a coherent summary sentence as a whole. [Structure of the Invention] (Means for Solving the Problems) In order to solve the above problems, the present invention divides each part included in one sentence by a summary control unit and specifies each divided part. The summarizing section instructs the summarizing section to perform summarization by the above-described method, and the summarizing section synthesizes and edits the summarizing result of each portion output by the summarizing section and outputs the result. Further, the summary range in one sentence can be specified by the user. (Operation) According to the present invention, first, the summary control unit operates on the sentence data input by the sentence input unit. The summarization control unit allows the user to specify the summarization range of the text with reference to the format and content of the text, and to specify the necessary summarization processing method for each part. Each divided part of the summary processing method (summary range)
Is input to the summarization processing unit together with information on the summarization processing method for each part, and the summarization processing unit summarizes the input part based on the specified information. The output of each digested section is sent to the digest editor,
The summary editing unit summarizes the summary results in the order of the input text or, for example, in the order specified, and sends the output data to the display unit as one summary text. Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of the text summarization system in the present embodiment. The input device 1 inputs a sentence to be summarized as character data. Character input data input from the input device 1 including the keyboard 101 and the magnetic recording medium driver 102 is input via the input control device 2. Is sent to the summary controller 3 as input text data. The digest control device 3 is connected to a digest instruction input device 4 as necessary. The summary instruction input device 4 divides the sentence data input to the summary control device 3 for each summary unit, and gives an instruction to give information of a summary processing method to each part. In this embodiment, the keyboard 40 is referred to while the input character data is referred to on the CRT display 401.
The method for the user to input instructions from 2 is adopted. The digest control device 3 divides the input sentence data into digest processing units using the instruction input from the digest instruction input device 4,
The information is sent to the summarization processing unit 5 together with the summarization processing method information from the summarization instruction input unit 4 for each processing unit. The summary processing device 5 summarizes the input individual summary processing units based on the information on the respective summary processing methods, and sends the result to the summary editing device 6. The summary editing device 6 edits each summary result sent from the summary processing device in the order of input, and outputs the result to the output device 8 via the output control device 7 as summary output text data.
The output device includes, for example, a CRT display 801, a printer 802, a magnetic recording medium driver 803, and the like. In realizing the configuration diagram of FIG. 1, the keyboard 101 of the input device, the keyboard 402 of the summary instruction input device,
The CRT display 401 of the summary instruction input device, the CRT display 801 of the output device, and the magnetic recording medium driver 102 of the input device and the magnetic recording medium driver 803 of the output device
Can be shared. FIG. 2 is a diagram showing an example of sentence data input to the summary processing apparatus and a summary processing instruction in this embodiment. The part shown in this figure is a part instructing to normally perform the summarization processing, a part instructing to output the input data without summarizing the data as it is, and a part instructing that the output is unnecessary. FIG. 3 is a flowchart showing a data processing process in the summary control device → summarization processing device → summary editing device in FIG. In step S1, the digest control device extracts a digest processing unit from input text data. This processing unit is clarified by the breaks in FIG. Subsequently, the digest processing apparatus checks the contents of the digest processing method in step S2, and if a normal digest (in FIG. 2) is instructed, proceeds to step S31.
If the output (in FIG. 2) without summarizing is instructed, the process proceeds to step S32, and if the output is unnecessary (in FIG. 2), the process proceeds to step S33. In step S31, using a method conventionally performed,
The input sentence data is appropriately summarized and output. In step S32, the input sentence data is directly used as output data. No output is made in S33. In the summary editing device, the output data of S31 to
Summarize in 4. Actually, the data is accumulated in the output buffer.
In step S5, it is checked whether or not the input sentence data still remains, that is, whether or not there is an unprocessed summary processing unit.If there is an unprocessed summary processing unit, the process returns to step S1 for a new summary processing unit. The processing up to step S4 is repeated. When all the input text data has been processed, the processing in FIG. 3 ends. FIG. 4 shows the input sentence data exemplified in FIG.
It is a figure which shows the example of the summary sentence processed by the abstract system shown in the figure, and output to the output device. The present invention is not limited to the above embodiment. For example, the division of the summary processing unit and the processing method for each unit may be performed by using a control code different from ordinary text characters in text data input from the input device, using a control code. Good. FIG. 5 shows the contents of the input sentence data in this case. Here, << A >>, << B >>, and << C >> are control codes indicating the start of a summary processing unit which is normally summarized, output without summarization, and need not be output, respectively. This data structure can be adopted as the internal sentence data structure at the time when the input information from the summary instruction input device is added in the above embodiment. Also, the order in which each summary processing unit is output to the output device is
It can also be entered as part of the summary instruction information. Sixth
The figure shows an example in which output order information is further added to the input information in FIG. Here, # 1, which was entered after << A >> and << B >>
The numbers # 2,... Are control codes indicating the output order.
FIG. 7 is a sentence obtained by summarizing the input data of FIG. 6 in this embodiment. FIG. 8 is a display screen view for the user to arbitrarily specify the range to be summarized. Here, it is assumed that the summary range designation switching means is added to the keyboard 402 (or the keyboard 101) in FIG. When performing summarization while switching the summarization range, this display screen is divided into a first display area for displaying summarization target sentences and a second display area for displaying summarization sentences, and the summarization range is changed. Since it is explicitly shown in the same display area, the user can easily determine in which summary range the summary has been made. In addition, since the summarization range can be switched, when the summarization in a large unit (for example, the whole text) fails, the summarization in a small unit (for example, a chapter or a paragraph) can be easily switched. In short, the present invention does not depart from the gist thereof,
Various modifications can be made. [Effects of the Invention] As described above, according to the present invention, when summarizing a general sentence in which various forms and contents are mixed, each of the parts having different forms and contents, and It is possible to give a detailed instruction on the summarization method for each of the ranges.
As a result, the entire summary result desired by the user, which is easy to understand, can be output at once, and a practical sentence summarization system can be realized.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a configuration of a text summary editing system according to one embodiment of the present invention, and FIG. 2 is a diagram showing text data input to the system and instructions given to the data. FIG. 3 is a schematic diagram showing an example of the summary instruction information, FIG. 3 is a flowchart showing the flow of text data in the processing of the summary control device / summary processing device / summary editing device of the system, and FIG. FIG. 5 is a schematic diagram showing an example of a sentence output to the output device as a result of summarizing the input sentence data of FIG. 5. FIG. 5 shows the data structure of a sentence obtained by additionally inputting the summary instruction information of FIG. 2 to the input sentence data of FIG. FIG. 6 is a schematic diagram showing the data structure of a sentence by further adding output order information to the sentence data of FIG. 5, and FIG. 7 is a diagram showing the data structure of the sentence data shown in FIG. Output device that summarizes input text data FIG. 8 is a schematic diagram showing an example of a sentence output to the device, and FIG. 8 is a schematic diagram showing an example of a sentence output to the output device as a result of adding the summary range designation switching means to the system shown in FIG. DESCRIPTION OF SYMBOLS 1 ... Input device 2 ... Input control device 3 ... Summary control device 4 ... Summary instruction input device 5 ... Summary processing device 6 ... Summary editing device 7 ... Output control device 8 ... Output apparatus

──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Yumiko Sugiura 1 Toshiba-cho, Komukai-shi, Kawasaki-shi Toshiba Research Institute, Inc. (72) Inventor Hisahiro Adachi Toshiba-cho, Komukai-ku, Kawasaki-shi 1 Toshiba Sogo In the laboratory (72) Inventor Kimito Takeda 1 Toshiba-cho, Komukai-shi, Kawasaki-shi Toshiba Research Institute, Inc. Inventor Shinya Amano 1 Toshiba-cho, Komukai, Sachi-ku, Kawasaki-shi Toshiba Research Institute, Inc. (56) Reference JP-A-62-109127 (JP, A) JP-A-60-138670 (JP, A) JP-A Sho 61-117658 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/21 G06F 17/30 G06F 17/27

Claims (1)

(57) [Claims] Means for designating a range in which summarization processing is to be performed among a plurality of input sentences, means for instructing a method of summarization processing for the specified range, and a method for summarizing the designated range. And a means for summarizing the sum, and summarizing the results and outputting the result as a summary sentence. 2. The specified summarization processing range is for the whole text,
The sentence summarizing system according to claim 1, wherein the sentence summarizing system is a chapter unit, a paragraph unit, or the like. 3. 2. The method according to claim 1, wherein the result summarized by the designated summarization method is one of a result obtained by a predetermined normal summarization process, a result not summarized, and a result not output. Sentence summarization system.