JP6213196B2 - Information processing program and information processing apparatus - Google Patents

Description

  The present invention relates to an information processing program and an information processing apparatus.

  As a conventional technique, an information processing apparatus that shapes a sentence including a plurality of polar expressions has been proposed (see, for example, Patent Document 1).

  The information processing device disclosed in Patent Literature 1 receives a situation sentence representing facts and an emotion word representing emotions as language data, and performs a morphological analysis on the situation sentence to decompose the situation sentence into words, The first emotion polarity and the second emotion polarity conflict with each other by determining the first emotion polarity from the emotion word and determining the second emotion polarity from the word obtained by decomposing the situation sentence. If it is, the sentence is shaped by deleting the word for which the second emotion polarity is determined from the situation sentence.

JP 2009-163565 A

  An object of the present invention is to provide an information processing program and an information processing apparatus that extract data that includes a polar expression uniformly from language data.

  In order to achieve the above object, one embodiment of the present invention provides the following information processing program and information processing apparatus.

[1] Connect the computer
Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Analyzing means for analyzing the polarity of a word included in the extracted data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
When the intensity of the analyzed polarity is larger than a predetermined value, the range of the extracted data is expanded. When a plurality of polarities are mixed in the extracted data, the extraction is performed in a range including each polarity. Changing means for dividing the extracted data and changing the range of the extracted data;
An information processing program for causing the changing means to function as output means for outputting data whose range has been changed as extracted data.

[2] Extraction means for extracting data in a predetermined range as a target for analyzing a computer from language data;
Analyzing means for analyzing the polarity of a word included in the extracted data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
When the intensity of the analyzed polarity is larger than a predetermined value, the range of the extracted data is expanded. When a plurality of polarities are mixed in the extracted data, the extraction is performed in a range including each polarity. Changing means for dividing the extracted data and changing the range of the extracted data;
An information processing program for functioning as teacher data output means for outputting data whose range has been changed by the changing means as teacher data used for machine learning.

[3]
Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Determination means for multi-value classification of the extracted data using a learning model prepared in advance;
Analyzing means for analyzing the polarity of a word included in the determined data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
If the strength of the analyzed polarity is greater than a predetermined value, the range of the determined data is expanded, and if a plurality of polarities are mixed in the determined data, the determination is made within a range including each polarity. Dividing the processed data and functioning as a changing means for changing the determined data range,
The determination means is an information processing program for performing multi-value classification on data whose range has been changed.

[4]
Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Determination means for multi-value classification of the extracted data using a learning model prepared in advance;
Analyzing means for analyzing the polarity of a word included in the determined data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
If the strength of the analyzed polarity is greater than a predetermined value, the range of the determined data is expanded, and if a plurality of polarities are mixed in the determined data, the determination is made within a range including each polarity. Changing means for dividing the determined data and changing the range of the determined data;
An information processing program for functioning as relearning means for relearning the learning model using the data in which the range is changed as teacher data.

[5] extraction means for extracting data in a predetermined range as an object to be analyzed from language data;
Analyzing means for analyzing the polarity of a word included in the extracted data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
When the intensity of the analyzed polarity is larger than a predetermined value, the range of the extracted data is expanded. When a plurality of polarities are mixed in the extracted data, the extraction is performed in a range including each polarity. Changing means for dividing the extracted data and changing the range of the extracted data;
An information processing apparatus comprising: output means for outputting the data whose range has been changed by the changing means as extracted data.

  According to the invention which concerns on Claim 1 or 5, the data in which polarity expression is contained uniformly can be extracted from language data.

  According to the second aspect of the present invention, the accuracy of multi-level classification can be improved by using the data including the polar expression uniformly as the teacher data of the multi-level classifier.

  According to the third aspect of the invention, it is possible to improve the accuracy of multi-level classification by performing multi-level classification on data in which polarity expressions are uniformly included.

  According to the fourth aspect of the present invention, it is possible to improve the accuracy of machine learning by performing relearning using data in which the polar expression is uniformly included.

FIG. 1 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the first embodiment. FIG. 2 is a schematic diagram illustrating an example of the configuration of the polarity expression dictionary data. FIGS. 3A to 3C are diagrams illustrating an example of a configuration of data extracted by the data extraction unit. FIGS. 4A to 4C are schematic diagrams illustrating the relationship between the data before the range change and the data whose range has been changed. FIG. 5 is a flowchart illustrating an example of the operation of the information processing apparatus. FIG. 6 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the second embodiment. FIG. 7 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the third embodiment. FIG. 8 is a flowchart illustrating an example of the operation of the information processing apparatus. FIG. 9 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the fourth embodiment. FIG. 10 is a flowchart illustrating an example of the operation of the information processing apparatus. FIGS. 11A to 11C are diagrams illustrating an example of a configuration of teacher data candidates extracted by the data extraction unit. FIG. 12 is a flowchart illustrating an example of the operation of the information processing apparatus.

[First Embodiment]
(Configuration of information processing device)
FIG. 1 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the first embodiment.

  As an example, the information processing apparatus 1 extracts data in units of sentences and clauses from language data 111 including a plurality of sentences, the polarity expression is uniformly included in the data, and the data is extracted data 113. Is output as Here, “polarity” has two or more values such as “positive” and “negative”, and “polarity expression” means “joyful”, “fun”, etc. in which the polarity belongs to “positive”. It shall indicate a morpheme or word such as “boring” or “get bored” whose polarity is “negative”.

  The information processing apparatus 1 includes a control unit 10 that includes a CPU and the like, controls each unit and executes various programs, and a storage device that includes a recording medium such as an HDD (Hard Disk Drive) and a flash memory and stores information. A storage unit 11 as an example and a communication unit 12 connected to an external device via a network (not shown) are provided.

  The control unit 10 functions as a language data acquisition unit 100, a data extraction unit 101, a polarity analysis unit 102, a data range change unit 103, an extracted data output unit 104, and the like by executing an information processing program 110 described later.

  The language data acquisition unit 100 acquires language data corresponding to a specified condition from the language data 111 in the storage unit 11. The conditions can be changed as appropriate. For example, the user who created the data, the date and time when the data was created, and the like are specified.

  The data extraction unit 101 extracts data in units determined in advance from the language data 111 acquired by the language data acquisition unit 100. Although the predetermined unit can be appropriately changed, in this embodiment, for example, a sentence is used as a unit. Data may be extracted in units of sentences within a predetermined number of characters.

  The polarity analysis unit 102 analyzes the type of polarity, the strength of the polarity, and the like included in the data by referring to the polarity expression dictionary data 112 described later for each of the data extracted by the data extraction unit 101. Here, the “polarity strength” means a degree indicating the degree in a plurality of stages (“3, 2, 1” and “large, medium, small”) as shown in FIG. “Yes, No”) may be included, and may be in accordance with other notation as long as it indicates strength.

  The data range changing unit 103 changes the range of data extracted by the data extracting unit 101 based on the analysis result of the polarity analyzing unit 102 by expanding or dividing the data range in units of sentences or clauses. Note that the data range changing unit 103 may change the data range only by expansion or division. In addition, when the data range is expanded, the data range changing unit 103 may extend in units of sentences or clauses that follow the data, or in units of sentences or clauses that precede the data. It may be extended. Further, it may be extended before and after the data.

  The extracted data output unit 104 outputs the data whose range has been changed by the data range changing unit 103 as the extracted data 113.

  The storage unit 11 stores an information processing program 110, language data 111, polarity expression dictionary data 112, extracted data 113, and the like.

  The information processing program 110 is a program that causes the control unit 10 to function as the above-described units 100 to 104 by being executed by the control unit 10.

  The language data 111 is, for example, Japanese and includes a plurality of sentences. Sentences are text information exchanged by e-mail, microblogs where text information is posted by multiple users, information obtained by converting voice into text, and information obtained by optically scanning printed paper. Etc. The language data 111 is not limited to Japanese, and other languages may be used. The language data 111 may be acquired from the outside.

  The polar expression dictionary data 112 is information in which words that are polar expressions are registered in advance together with the type of polarity and the strength of the polarity.

  The extracted data 113 is data extracted by the extracted data output unit 104.

  The information processing device 1 is, for example, a server device or a personal computer, and a mobile phone or a mobile information processing terminal can be used.

  FIG. 2 is a schematic diagram illustrating an example of the configuration of the polarity expression dictionary data 112.

  The polarity expression dictionary data 112a is an example of the polarity expression dictionary data 112, and includes a word that is a polarity expression, a polarity that is a type of polarity of the word, and a strength of the polarity. For example, the word “fun” has a polarity of “positive” and a polarity of “2”.

(Operation of information processing device)
Next, the operation of the first embodiment will be described.

  FIG. 5 is a flowchart illustrating an example of the operation of the information processing apparatus 1.

  First, the language data acquisition unit 100 acquires the language data 111 in which the conditions specified from the language data 111 in the storage unit 11, for example, the user who created the data, the date and time when the data was created, and the like meet the conditions ( S1).

  Next, the data extraction unit 101 extracts data from the language data 111 acquired by the language data acquisition unit 100 in units of sentences within 35 characters, for example (S2). An example of the configuration of data extracted according to the conditions is shown below.

  FIGS. 3A to 3C are diagrams illustrating an example of a configuration of data extracted by the data extraction unit 101. FIG.

  As shown in FIG. 3A, the data 101a is composed of two sentences and is 21 characters. Further, as shown in FIG. 3B, the data 101b is one sentence and is 32 characters. Further, as shown in FIG. 3C, the data 101c is one sentence and is 21 characters.

  Next, the polarity analysis unit 102 refers to the polarity expression dictionary data 112 for each of the data 101a-101c extracted by the data extraction unit 101, thereby determining the type of polarity, the strength of the polarity, etc. included in the data. Analyze (S3, S4, S7).

  For example, since the polarity analysis means 102 includes the words “happy” and “happy” in the data 101a, the polarity of the data 101a is “positive” and the strength of the polarity is “5”. It is determined that the polarity is strong (S4; Yes). It is assumed that the polarity determined to be strong is, for example, “5” or more.

  In addition, since the data 101b includes the words “fun” and “tired”, the polarity analysis unit 102 determines that the polarity of the data 101b includes both “positive” and “negative”. (S4; No, S7; Yes).

  Further, since the polarity analysis means 102 includes “interesting” in the data 101c, the polarity of the data 101c is “positive”, the strength of the polarity is “1”, and the polarity is not strong (S4). ; No), it is determined that the polarities are not mixed (S7; No).

  Next, the data range changing unit 103 changes the range of the data 101a-101c based on the analysis result of the polarity analyzing unit 102 as described below.

  4A to 4C are schematic diagrams illustrating the relationship between the data 101a to 101c before the range change and the data whose range has been changed.

Data range changing unit 103, the data 101a described above more polar, statements that follows also determines that there is a high probability that contains the same polarity, the data 103a ₂ is a sentence that follows the same data 103a ₁ and data 101a The range is expanded to the data 103a after the range change (S5).

The data range changing means 103, data 101b described above because the polarities are mixed, divides the data 101a to the data 103b ₁ range including respective polarity and the data 103b ₂ (S8).

  Further, the data range changing unit 103 determines that the data 101c described above is data that does not need to be changed because the polarity of the data 101c is not strong and the polarity is not mixed, and maintains the range of the data 101c. Data 103c is set (S9).

Next, the extracted data output means 104 outputs the data 103a, 103b ₁ , 103b ₂ , 103c whose ranges have been changed by the data range changing means 103 as the extracted data 113 and stores them in the storage unit 11 (S6).

(Effects of the first embodiment)
According to the first embodiment described above, the polarity of the data extracted by the polarity analysis unit 102 is analyzed. When the polarity is strong, the range is expanded, and when a plurality of polarities are included, the data is divided. Can be extracted.

  The data range changing unit 103 may repeat steps S3-S5 and S7-S9 a plurality of times.

[Second Embodiment]
(Configuration of information processing device)
The second embodiment is an application of the first embodiment, and differs in that the extracted data 113 is output as teacher data used for machine learning.

  FIG. 6 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the second embodiment. In addition, the same code | symbol is attached | subjected about the structure which has the same function as 1st Embodiment.

  The control unit 10 of the information processing apparatus 1A functions as the teacher data output unit 105 instead of the extracted data output unit 104 by executing the information processing program 110A.

  The teacher data output means 105 obtains data by means 100-103 functioning in the same manner as in the first embodiment, and outputs it as teacher data 114.

  The storage unit 11 of the information processing apparatus 1 </ b> A has teacher data 114 instead of the extracted data 113.

  The teacher data 114 is collected as information used for machine learning, and is transmitted to an external machine learning device (not shown) via the communication unit 12. Learning is performed using a method such as support vector machine, naive bayes, adaboost, decision tree and the like.

(Effect of the second embodiment)
According to the second embodiment described above, the information processing apparatus 1 of the first embodiment can be applied to the collection of the teacher data 114, and since the teacher data 114 includes the polar expression uniformly, noise is generated. The information that can be used as features is decreased, and as a result, when the teacher data 114 is used for machine learning, the accuracy of learning is improved.

[Third Embodiment]
(Configuration of information processing device)
The third embodiment is an example in which the first embodiment is applied to a multilevel classifier.

  FIG. 7 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the third embodiment. In addition, the same code | symbol is attached | subjected about the structure which has the function similar to 1st Embodiment and 2nd Embodiment.

  The control unit 10 of the information processing apparatus 1B further functions as the multi-value classification determination unit 106 by executing the information processing program 110B, and functions as the determination result output unit 107 instead of the extracted data output unit 104.

  The multi-value classification determination unit 106 determines multi-value classification for data given based on a learning model 115 described later. Here, the multi-value classification determination is to determine which given data is classified into a plurality of classes prepared in advance. For example, when a microblog is used as data, “positive” is determined. , “Negative”, “Neutral”, categorized as ternary polarity, or “data”, “packaging”, “price”, “ It is classified into values such as “design” and “other”.

  The determination result output unit 107 associates the data to be determined with the class classified by the multi-value classification determination unit 106 and outputs the result as determination result data 116.

  The storage unit 11 of the information processing apparatus 1 </ b> B further includes a learning model 115 and includes determination result data 116 instead of the extracted data 113.

  The learning model 115 is generated by machine learning and is used for the multi-value classification determination unit 106.

  The determination result data 116 is data output by the determination result output unit 107.

(Operation of information processing device)
Next, the operation of the third embodiment will be described.

  FIG. 8 is a flowchart illustrating an example of the operation of the information processing apparatus 1B.

  First, the language data acquisition unit 100 acquires the language data 111 in which the conditions specified from the language data 111 in the storage unit 11, for example, the user who created the data, the date and time when the data was created, and the like meet the conditions ( S10).

  Next, the data extraction unit 101 extracts data from the language data 111 acquired by the language data acquisition unit 100 in units of sentences within 35 characters, for example (S11).

  Next, the multi-level classification determination unit 106 determines multi-level classification based on the learning model 115 for the data extracted by the data extraction unit 101 (S12).

  Next, the polarity analysis unit 102 refers to the polarity expression dictionary data 112 for each of the data that the multi-level classification determination unit 106 has determined multi-level classification, so that the type of polarity and the strength of the polarity included in the data are determined. Etc. are analyzed (S13, S14, S16).

  Next, when the polarity of the data subjected to the multi-value classification determination is strong (S14; Yes), the data range changing unit 103 extends the range to the subsequent sentence data and changes the range (S15). In addition, when the determination result of the multi-value classification determination unit 106 has a high reliability, it may not be expanded in consideration of the reliability.

  Further, when the polarity is mixed in the data determined by the multi-value classification (S14; No, S16; Yes), the data range changing unit 103 divides the data into the data including the respective polarities of the data determined by the multi-value classification. (S17). In addition, when the determination result of the multi-value classification determination unit 106 has a high reliability, it may not be divided in consideration of the reliability.

  Further, the data range changing means 103 is data that does not need to be changed in the extraction range when the polarity of the data subjected to the multi-value classification determination is not strong (S14; No) and the polarities are not mixed (S16; No). Judgment is made and the range of data is maintained (S18).

  Next, the multi-level classification determination unit 106 determines the multi-level classification of the data whose range has been changed by the data range change unit 103 based on the learning model 115 (S19).

  Next, the determination result output unit 107 outputs the determination result data 116 by associating the data targeted by the multilevel classification determination unit 106 with the class classified by the multilevel classification determination unit 106 (S20).

(Effect of the third embodiment)
According to the third embodiment described above, the information processing apparatus 1 of the first embodiment can be applied to a multi-level classifier. Specifically, the reliability of the multi-level classification and the polarity analysis The classification target can be analyzed from both viewpoints, and the accuracy of the multi-level classification can be improved by performing multi-level classification on the data in which the polar expression is uniformly included.

[Fourth Embodiment]
In the fourth embodiment, by applying the first embodiment to the data determined by multilevel classification using a learning model prepared in advance, the range is changed so that the polar expression is uniformly included. It is used as teacher data, and the learning model is relearned using the teacher data (application to a bootstrap method).

  FIG. 9 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the fourth embodiment. In addition, the same code | symbol is attached | subjected about the structure which has the function similar to 1st Embodiment-3rd Embodiment.

  The control unit 10 of the information processing apparatus 1C further functions as the learning model relearning unit 108 by executing the information processing program 110C.

  The learning model relearning unit 108 relearns the learning model 115 using the teacher data 114.

(Operation of information processing device)
Next, the operation of the fourth embodiment will be described.

  FIG. 10 is a flowchart illustrating an example of the operation of the information processing apparatus 1C.

  First, the language data acquisition unit 100 acquires language data 111 corresponding to conditions specified from the language data 111 in the storage unit 11, for example, conditions such as the user who created the data and the date and time when the data was created ( S30).

  Next, the data extraction unit 101 extracts data from the language data 111 acquired by the language data acquisition unit 100 in units of sentences within 35 characters, for example (S31).

  Next, the multi-level classification determination unit 106 determines multi-level classification for the data extracted by the data extraction unit 101 based on the learning model 115 (S32).

  Next, the polarity analysis unit 102 refers to the polarity expression dictionary data 112 for each of the data that the multi-level classification determination unit 106 has determined multi-level classification, so that the type of polarity and the strength of the polarity included in the data are determined. Etc. are analyzed (S33, S34, S36).

  Next, when the polarity of the data subjected to the multi-value classification determination is strong (S34; Yes), the data range changing unit 103 extends the range to the data that is a subsequent sentence and changes the range (S35). In addition, when the determination result of the multi-value classification determination unit 106 has a high reliability, it may not be expanded in consideration of the reliability.

  Further, when the polarity is mixed in the data determined by the multi-level classification (S34; No, S36; Yes), the data range changing unit 103 divides the data into the range including the respective polarities of the data determined by the multi-level classification. (S37). In addition, when the determination result of the multi-value classification determination unit 106 has a high reliability, it may not be divided in consideration of the reliability.

  Further, the data range changing means 103 is data that does not need to be changed in the extraction range when the polarity of the data subjected to the multi-value classification determination is not strong (S34; No) and the polarities are not mixed (S36; No). Judgment is made and the range of data is maintained (S38).

  Next, the teacher data output means 105 outputs the data whose range has been changed by the data range change means 103 as teacher data 114 (S39).

  Next, the learning model re-learning means 108 re-learns the learning model 115 using the teacher data 114 (S40).

  When the result determined by the multi-value classification determination unit 106 is not used as the relearning teacher data 114, the determination result output unit 107 outputs the determination result data 116 as in the third embodiment. To do.

(Effect of the fourth embodiment)
According to the above-described fourth embodiment, the information processing apparatus 1 according to the first embodiment can be applied to relearning of a learning model, and relearning is performed using data that includes a polar expression uniformly. Thus, the accuracy of machine learning can be improved.

(Example)
In this embodiment, a microblog such as “twitter” (registered trademark) is used as the language data 111 in the second embodiment, and the teacher data 114 is extracted from the microblog. Further, as a polarity determination, it is determined whether the polarity is “positive” or “negative”.

  FIG. 12 is a flowchart illustrating an example of the operation of the information processing apparatus 1A.

  First, the language data acquisition unit 100 acquires language data 111 corresponding to conditions specified from the language data 111 in the storage unit 11, for example, conditions such as the user who created the data and the date and time when the data was created ( S50).

  Next, the data extraction unit 101 extracts teacher data candidates from the language data 111 acquired by the language data acquisition unit 100, for example, in units of one article (one “tweet” in “twitter”) (S51). ). An example of the configuration of data extracted according to the conditions is shown below.

  FIGS. 11A to 11C are diagrams illustrating an example of a configuration of teacher data candidates extracted by the data extraction unit 101. FIG.

  As shown in FIG. 11A, the teacher data candidate 101d consists of two sentences. As shown in FIG. 11B, the teacher data candidate 101e is one sentence. In addition, as shown in FIG. 11C, the teacher data candidate 101f is one sentence.

  Next, the polarity analysis means 102 analyzes the type of polarity expression, the strength of polarity, etc. included in the data by referring to the polarity expression dictionary data 112 for each of the teacher data candidates 101d-101f (S52). ).

  For example, the polarity analysis unit 102 determines that the teacher data candidate 101d is “positive” because the teacher data candidate 101d includes the words “completed” and “achievement”, and the polarity is very strong. (S53; Yes).

  Further, since the polarity analysis means 102 includes the words “interesting” and “tired” in the teacher data candidate 101e, the polarity of the teacher data candidate 101e is contradictory between “positive” and “negative”. Judgment is made (S53; No, S55; Yes).

  In addition, since the polarity analyzing unit 102 includes “delicious” in the teacher data candidate 101f, the polarity of the teacher data candidate 101f is “positive”, but the polarity is not strong (S53; No), and the polarity is It is determined that there is no conflict (S55; No).

  Next, the data range changing unit 103 changes the range of the teacher data candidates 101d to 101f based on the analysis result of the polarity analyzing unit 102 as described below.

  The data range changing unit 103 determines that the above-described teacher data candidate 101d has a strong polarity, so that “reply” and “retweet” that are replies have a high probability of including the same polarity, and adds the teacher data candidate 103d ( S54).

Further, the data range changing unit 103 discards the teacher data candidate 103e ₁ that is a reply “reply” or “retweet” because the teacher data candidate 101e described above has a mixed polarity (S56), and the teacher data dividing the candidate 101e to each of the polar "positive" and scope teacher data candidate 103e ₂ and the teacher data candidates 103e _3, including a "negative" (S57).

  Further, the data range changing unit 103 determines that the above-described teacher data candidate 101f is data that does not require a change of the extraction range because the polarity is not strong and the polarity is not mixed, and the range of the teacher data candidate 101f is determined. And the teacher data candidate 103f that is a reply “reply” or “retweet” is discarded (S58).

Then, the extracted data output unit 104 outputs the data 101d scope by data range changing unit 103 is changed and 103d, the 103e ₂ and 103e _3, 101f as teacher data 114, stored in the storage unit 11 (S59) .

[Other embodiments]
The present invention is not limited to the above embodiment, and various modifications can be made without departing from the spirit of the present invention.

  In the above embodiment, the function of each means 100-108 in the control unit 10 is realized by a program, but all or a part of each means may be realized by hardware such as ASIC. The program used in the above embodiment can be provided by being stored in a recording medium such as a CD-ROM. In addition, replacement, deletion, addition, and the like of the above-described steps described in the above embodiment are possible within a range that does not change the gist of the present invention.

1, 1A, 1B, 1C Information processing apparatus 10 Control unit 11 Storage unit 12 Communication unit 100 Language data acquisition unit 101 Data extraction unit 102 Polarity analysis unit 103 Data range change unit 104 Extracted data output unit 105 Teacher data output unit 106 Multi-value Classification determination means 107 Determination result output means 108 Learning model relearning means 110 Information processing program 110A-110C Information processing program 111 Language data 112 Polarity expression dictionary data 113 Extraction data 114 Teacher data 115 Learning model 116 Determination result data

Claims (5)

Computer
Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Analyzing means for analyzing the polarity of a word included in the extracted data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
When the intensity of the analyzed polarity is larger than a predetermined value, the range of the extracted data is expanded. When a plurality of polarities are mixed in the extracted data, the extraction is performed in a range including each polarity. Changing means for dividing the extracted data and changing the range of the extracted data;
An information processing program for causing the changing means to function as output means for outputting data whose range has been changed as extracted data. Extracting means for extracting data in a predetermined range as a target for analyzing a computer from language data;
Analyzing means for analyzing the polarity of a word included in the extracted data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
When the intensity of the analyzed polarity is larger than a predetermined value, the range of the extracted data is expanded. When a plurality of polarities are mixed in the extracted data, the extraction is performed in a range including each polarity. Changing means for dividing the extracted data and changing the range of the extracted data;
An information processing program for functioning as teacher data output means for outputting data whose range has been changed by the changing means as teacher data used for machine learning. Computer
Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Determination means for multi-value classification of the extracted data using a learning model prepared in advance;
Analyzing means for analyzing the polarity of a word included in the determined data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
If the strength of the analyzed polarity is greater than a predetermined value, the range of the determined data is expanded, and if a plurality of polarities are mixed in the determined data, the determination is made within a range including each polarity. Dividing the processed data and functioning as a changing means for changing the determined data range,
The determination means is an information processing program for performing multi-value classification on data whose range has been changed. Computer
Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Determination means for multi-value classification of the extracted data using a learning model prepared in advance;
Analyzing means for analyzing the polarity of a word included in the determined data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
If the strength of the analyzed polarity is greater than a predetermined value, the range of the determined data is expanded, and if a plurality of polarities are mixed in the determined data, the determination is made within a range including each polarity. Changing means for dividing the determined data and changing the range of the determined data;
An information processing program for functioning as relearning means for relearning the learning model using the data in which the range is changed as teacher data. Extracting means for extracting data in a predetermined range as an object to be analyzed from language data;
Analyzing means for analyzing the polarity of a word included in the extracted data based on a polarity expression dictionary that associates a word with polarity and polarity strength;
When the intensity of the analyzed polarity is greater than a predetermined value, the range of the extracted data is expanded. Changing means for dividing the extracted data and changing the range of the extracted data;
An information processing apparatus comprising: output means for outputting the data whose range has been changed by the changing means as extracted data.

Images