Information providing system

Abstract

To provide an information providing system configured to exclude a posted sentence including a negative word from a display target.SOLUTION: An information providing system is configured to: acquire a posted sentence posted on a network together with spot information that specifying a spot associated with the posted sentence; analyze the acquired posted sentence to predict post emotion which is the emotion of a poster who has posted the posted sentence; aggregate the predicted post emotions, for each spot associated with a posted sentence from which the post emotion is analyzed; and specify the post emotion for each spot. The system is further configured to display, on a screen, at least a part of the posted sentence in addition to the specified post emotion, while displaying the posted sentences excluding a posted sentence including a negative word.SELECTED DRAWING: Figure 12

Description

Technical Field

[0001] The present invention relates to an information providing system that provides information based on a posted text posted on a network.

Background Art

[0002] In recent years, systems have been provided that allow users to post posted texts on a computer network via a terminal and view posted texts posted by other users. For example, there are blogs, SNS (Social Network Service), X (registered trademark), chats, etc. (hereinafter referred to as SNS, etc.). In addition, many of these posted texts include posted texts that post information about locations such as facilities visited by the poster.

[0003] Here, the above SNS, etc. have the merit of being able to quickly obtain the latest information experienced or felt by users who have actually visited the site, and various systems for providing information based on posted texts posted on SNS, etc. have been proposed. For example, Japanese Patent Application Laid-Open No. 2019-20784 collects posted texts posted on SNS and information for specifying the posting location, analyzes the posted texts posted at the same location, and determines whether positive feelings or negative feelings are dominant among the posters who posted the posted texts for that location as a whole. A technique has been proposed to display the determination result as an emotion mark in a map image and also display the comments of the posted text.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Here, in Patent Document 1 described above, for a posted text posted on SNS or the like, not only the feelings of the poster but also the posted text itself that is the source of the analysis of the feelings are displayed. However, for a posted text with a negative feeling in particular, it may include negative, aggressive, or discriminatory expressions, and it is desirable to avoid ethically making them display targets.

[0006] The present invention has been made to solve the above-mentioned conventional problems, and while making a posted text and posted feelings obtained by analyzing the posted text display targets, it is possible to exclude a posted text including a negative word from the display targets, and an object of the present invention is to provide an information providing system.

Means for Solving the Problem

[0007] To achieve the above object, an information providing system according to the present invention includes: a posted text information acquisition means for acquiring a posted text posted on a network together with location information for specifying a location associated with the posted text; a feeling prediction means for predicting a posted feeling, which is the feeling of a poster who posted the posted text, by analyzing the posted text acquired by the posted text information acquisition means; a location feeling specifying means for classifying and aggregating the posted feelings predicted by the feeling prediction means for each location associated with the posted text that is the source of the analysis, and specifying the posted feelings for each location; and a posted feeling display means for displaying at least a part of the posted text associated with the specified location on a screen in addition to the posted feelings at the specified location. The posted feeling display means excludes and displays a posted text including a negative word among the posted texts associated with the specified location. Note that the "negative word" is not limited to a word that gives a negative impression, and also includes other words that give discomfort, anger, or sadness when viewed by a third party. For example, it also includes words of aggressive expressions, discriminatory expressions, and slang.

Effect of the Invention

[0008] According to the information providing system of the present invention having the above configuration, while making the post text and the post sentiment obtained by analyzing the post text the display targets, it is possible to exclude from the display targets the post texts including negative words that are not desirable for third parties to view.

Brief Description of Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Embodiments for Carrying Out the Invention

[0010] Hereinafter, a detailed description will be given with reference to the drawings based on an embodiment in which the information providing system according to the present invention is embodied. First, the schematic configuration of the information providing system 1 according to the present embodiment will be described with reference to FIG. 1. FIG. 1 is a schematic configuration diagram showing the information providing system 1 according to the present embodiment.

[0011] As shown in FIG. 1, the information providing system 1 according to the present embodiment basically includes an information providing server 3 provided in the information providing center 2 and a communication terminal 5 possessed by the user 4. Further, the information providing server 3 and the communication terminal 5 are configured to be able to transmit and receive electronic data to and from each other via the communication network 6. Note that examples of the communication terminal 5 include a mobile phone, a smartphone, a tablet terminal, a personal computer, and a navigation device which is an in-vehicle device.

[0012] Here, the information providing server 3 is a server device that manages information to be provided to the communication terminal 5 (that is, the user who possesses the communication terminal 5). The information providing server 3 stores information about information providing points across the country to be provided to the communication terminal 5 in the distribution information DB 7. Note that the information providing points are not particularly limited in genre or scale, and examples include commercial facilities such as restaurants and retail stores, public facilities such as stations and hospitals, and other accommodation facilities, parking lots, etc. Further, it is not limited to facilities, and for example, tourist spots may be used. In addition, as information about the information providing points stored in the distribution information DB 7 as described later, in addition to the name, address, position coordinates, telephone number, business hours, etc. of the information providing points, in the present embodiment, in particular, "posted texts" posted on the network for the information providing points and information regarding "posted feelings" identified by analyzing the posted texts are included. Details regarding the posted texts and posted feelings will be described later. Then, the information providing server 3 provides (distributes) the information about the information providing points stored in the DB to the communication terminal 5 via the communication network 6.

[0013] In addition, the communication terminal 5 is an information terminal owned by a user and equipped with communication functions, navigation functions, etc. Examples include mobile phones, smartphones, tablet terminals, personal computers, and navigation devices such as in-vehicle devices. In particular, when the communication terminal 5 is a terminal capable of executing applications such as a smartphone, an application program is installed that can display a map image of an area designated by the user as one of the applications, and can also display an icon indicating the posted emotion with respect to the position of the information-providing point included in the displayed map image. Further, in the above application program, by selecting the icon displayed on the map image, more detailed information regarding the information-providing point corresponding to the selected icon can be obtained from the information-providing server 3 and guided. Note that the selection of the information-providing point to be guided can be made using the above map image, or can also be made from the search results obtained by inputting search conditions. Also, the function of guiding regarding these information-providing points may be a part of the navigation function for guiding the movement to the destination, or may be executed by an application program different from the navigation function.

[0014] In addition, the communication network 6 includes a large number of base stations arranged throughout the country and a communication company that manages and controls each base station, and is configured by connecting the base stations and the communication company to each other by wire (optical fiber, ISDN, etc.) or wirelessly. Here, the base station has a transceiver (transmitter / receiver) and an antenna for communicating with the communication terminal 5. Then, while the base station performs wireless communication between communication companies, it becomes the end of the communication network 6 and has the role of relaying the communication of the communication terminal 5 within the range (cell) where the radio wave of the base station reaches with the information-providing server 3.

[0015] On the one hand, the information providing server 3 included in the above information providing system 1 can obtain a posted text on the network from a social networking service (hereinafter referred to as SNS) providing SNS server 8 on the network, together with location information specifying the location associated with the posted text and time information specifying the posted time (collectively referred to as posted text information). Note that the method of obtaining the posted text information may be through the network or through a storage medium such as a flash memory.

[0016] Here, the SNS server 8 is a server device that constructs an SNS on the network. The SNS is a community-type service for constructing connections with others (which may be not only individuals but also corporations). Service users can access from clients such as smartphones, tablet terminals, personal computers, etc., and by posting messages, images, etc., the content can be viewed by other service users. Also, other service users who have viewed the message or image can not only comment on the post, but also quote and retweet (repost) the content, follow the poster through the follow function, or give a positive reaction (e.g., 'Good', 'Like') when they empathize with the posted content. Note that in this embodiment, it is an open-type SNS that can be used by anyone as long as they are registered users without any particular restrictions on the users. Also, the SNS server 8 is provided with a storage DB9, and messages, image data, date and time of posting, hashtag, location where the posting was made (which may be a facility name or a location coordinate), number of positive reactions and number of retweets received from other service users for each post, number of followers for each service user, etc. are stored in the storage DB9.

[0017] Then, in the present embodiment, the information providing server 3 acquires, as post information, from the SNS server 8, among the data stored in the storage DB 9, in particular, the combination of the content of the post (which may be only the message, or may include the image if an image is attached), the date and time when the post was made, and the location where the post was made.

[0018] Subsequently, the configuration of the information providing server 3 in the information providing system 1 will be described in more detail with reference to FIG. 2. As shown in FIG. 2, the information providing server 3 includes a server control unit 11, a post information DB 13 as an information recording means connected to the server control unit 11, a distribution information DB 7, a map information DB 14, a dictionary DB 15, and a server-side communication device 16.

[0019] The server control unit 11 is a control unit (such as an MCU or an MPU) that controls the entire information providing server 3, and includes a CPU 21 as an arithmetic unit and a control unit, a RAM 22 used as a working memory when the CPU 21 performs various arithmetic processes, a ROM 23 in which, in addition to control programs, an emotion analysis processing program (FIG. 6) and an information providing processing program (FIG. 10) described later are recorded, and an internal storage device such as a flash memory 24 that stores the programs read from the ROM 23. Note that the server control unit 11 has various means as processing algorithms. For example, the post information acquisition means acquires a post posted on the network together with location information that identifies the location associated with the post. The emotion prediction means predicts the post emotion, which is the emotion of the poster who posted the post, by analyzing the post acquired by the post information acquisition means. The location emotion specifying means classifies and aggregates the post emotions predicted by the emotion prediction means for each location associated with the post that is the analysis source, and specifies the post emotion for each location.

[0020] The posted text information DB 13 is a storage means for storing the posted text information acquired from the SNS server 8. Here, the posted text information includes the posted text posted on the network as described above, the location information for identifying the location associated with the posted text, and the time information for identifying the posted time. Further, in particular, in this embodiment, as will be described later, it is determined whether or not the posted text information acquired from the SNS server 8 includes a negative word in the posted text, and the posted text determined to include a negative word is excluded from the display target for the user. Therefore, the posted text information DB 13 also stores a display target flag, which is a flag for identifying whether or not to exclude the posted text from the display target for each posted text information. Note that the "location associated with the posted text" is the location where the posted text was posted (the location where the poster who posted the posted text is located) in this embodiment. However, for example, when the posted text includes a place name, or when the place name is associated as a hashtag, the place name may be used as the location associated with the posted text.

[0021] Here, FIG. 3 is a diagram showing an example of the posted text information stored in the posted text information DB 13. As shown in FIG. 3, in the posted text information DB 13, the content of the posted text (text data of the message) is stored in association with the location name indicating the location linked to the posted text, the posted date and time, and the display target flag. Note that the display target flag is added not only to indicate whether to exclude the posted text from the display target, but also to identify what kind of negative words of emotions the posted text specifically contains, that is, information about the types of negative words. In the example shown in FIG. 3, it is stored separately for each location name linked to the posted text, but how to classify and store it can be changed as appropriate. Also, in the example shown in FIG. 3, the content of the posted text is only text information, but if an image is attached, the image may also be included. Also, as information indicating the location linked to the posted text, a location name is stored, but it may be position coordinates instead of the location name. Note that when the SNS server 8 receives a posted text posted on the network by a service user (poster), the position coordinates of the terminal used by the service user for posting (identified by GPS or the like provided in the terminal) are also acquired as the place where the posted text was posted. Therefore, when storing the specific location name where the posted text was posted as posted text information as shown in FIG. 3, it is necessary to refer to the map information and specify the location name of the location where the service user is predicted to be located from the acquired position coordinates, but that process may be performed by the SNS server 8 or the information providing server 3. Also, for the posted date and time, only the time may be stored, or on the other hand, the day of the week and the like may also be included.

[0022] Also, as described above, the distribution information DB 7 is a storage means for storing various information related to the information providing locations targeted for information provision across the country. Here, FIG. 4 is a diagram showing an example of the information stored in the distribution information DB 7.

[0023] As shown in FIG. 4, in the distribution information DB7, for information-providing locations across the country, location ID, location name, location coordinates, detailed location information, current congestion status, posted sentiment, posted text, etc. are stored. However, it is not always necessary to store all of these pieces of information in the distribution information DB7. Note that "posted sentiment" specifically identifies which sentiment the sentiment of the poster (more precisely, the poster who posted the posted text at that location) regarding the posted text about that location as a whole falls into, and is identified by the server control unit 11 by analyzing the posted text information stored in the posted text information DB13 as described later. Note that there are various human emotions, but in particular, in this embodiment, the posted sentiment is identified using eight basic emotions classified broadly. Furthermore, using two types, positive emotion (positive sentiment) and negative emotion (negative sentiment), as judgment factors, it is also to be determined whether the sentiment of the poster is closer to positive emotion or negative emotion, or whether it is neutral (neither) and not leaning towards either. On the other hand, regarding "posted text", the entire posted text may be stored, but in this embodiment, only the keywords indicating the evaluation and status of the information-providing location are extracted and stored from among the posted texts posted for the information-providing location. However, posted texts determined to include negative words are excluded from the display targets.

[0024] For example, in the distribution information DB7 shown in FIG. 4, for "XX Station" at the position coordinates (x1, y1), facility information, current congestion status, as well as posted sentiment and posted text are stored. Similarly, information regarding other information-providing locations is also stored.

[0025] Also, the map information DB14 is a storage means for storing map information. The map information is composed of various types of information necessary for route search, route guidance, and map display, including a road network. For example, it consists of link data regarding roads (links), node data regarding node points, intersection data regarding each intersection, location data regarding locations such as facilities, map display data for displaying the map, search data for searching for routes, search data for searching for locations, etc.

[0026] Then, the server control unit 11 uses the map information DB 14 to, for example, transmit map display data for displaying a map image on the communication terminal 5 in response to a request from the communication terminal 5, search for a point corresponding to the input search condition, or perform a route search from the departure point to the destination using the map information stored in the map information DB 14 when a route search request is received.

[0027] However, when the communication terminal 5 has map information, the communication terminal 5 can perform the above processing using the map information it has. In that case, the map information DB 14 is not necessarily required in the information providing server 3.

[0028] Also, the dictionary DB 15 is a storage means in which pre-prepared negative words are registered. Note that the negative words are not limited to words that give a negative impression, but also include other words that give discomfort, anger, or sadness when viewed by a third party. For example, it includes words of aggressive or discriminatory expressions, slang, etc. Furthermore, it is not limited to meaningful character strings, but also includes emoticons, emojis, ASCII art, etc. Also, it is not limited to nouns, but also includes adverbs, adjectives, etc., and the negative words are stored together with the part of speech of the word.

[0029] Then, the server control unit 11 determines whether or not the posted text information acquired from the SNS server 8 using the dictionary DB 15 contains a negative word in the posted text, that is, whether or not to exclude it from the display target to other users. Also, in this embodiment, in addition to the determination using the dictionary DB 15, determination using machine learning is further performed. Details will be described later.

[0030] On one hand, the server-side communication device 16 is a communication device for communicating with the communication terminal 5 which is the target of information transmission and reception via the communication network 6. In addition to the communication terminal 5, it is also possible to receive traffic information composed of various information such as traffic jam information, regulation information, traffic accident information, etc. transmitted from the Internet, traffic information centers, for example, VICS (registered trademark: Vehicle Information and Communication System) centers, etc. Furthermore, in addition to traffic information, by communicating with external servers, it is also possible to receive weather information in various regions across the country, event information regarding events held across the country, news in various places, congestion information at locations, etc.

[0031] Next, the schematic configuration of the communication terminal 5 owned by the user will be described with reference to FIG. 5. FIG. 5 is a block diagram schematically showing the control system of the communication terminal 5 according to the present embodiment. Hereinafter, the case where the communication terminal 5 is a smartphone will be described as an example.

[0032] As shown in FIG. 5, the communication terminal 5 is connected to a data bus BUS, a CPU 31, a memory 32 in which user information (user ID, name, etc.) regarding the user who owns the communication terminal 5 and application programs are stored, an input / output unit 35 which is an interface such as a microphone 33 and a speaker 34, a display 36 composed of a liquid crystal display panel, etc., an input operation unit 37 composed of a touch panel, a keyboard, etc., a GPS 38, and a transmission / reception circuit unit (RF) 39 for transmitting and receiving signals to and from the base station of the communication network 6.

[0033] Here, the CPU 31 built into the communication terminal 5 is a control means of the communication terminal 5 that executes various operations according to the operation program stored in the memory 32, and constitutes the communication terminal control unit 41 together with the memory 32. Also, the various processing contents of the communication terminal control unit 41 are displayed on the display 36 as necessary. Note that the communication terminal control unit 41 constitutes various means as a processing algorithm. For example, the post emotion display means displays at least a part of the post text linked to the designated location in addition to the post emotion at the designated location on the screen.

[0034] In addition to making calls, the communication terminal 5 can communicate via the transmission / reception circuit unit 39 to perform Internet communication, receive information on charging facilities from the information providing server 3, and receive traffic information consisting of various information such as traffic jam information, regulation information, and traffic accident information transmitted from a traffic information center, for example, a VICS (registered trademark) center or a probe center.

[0035] In addition, the memory 32 is a storage medium that stores user information (user ID, name, etc.) regarding the user who owns the communication terminal 5, map information, the user's browsing history of the web by the user, the user's movement history which is a history of position information detected based on the GPS 38 and other sensors, schedule information, etc. Also, various application programs including the information providing processing program (Figure 10) described later are stored. The memory 32 may be constituted by a hard disk, a memory card, etc.

[0036] In addition to voice output for calls, the speaker 34 outputs voice guidance for guiding driving along the guidance route (the user's planned movement route) based on an instruction from the communication terminal control unit 41 when the navigation function is executed.

[0037] In addition, the display 36 is disposed on one surface of the housing, and a liquid crystal display, an organic EL display, or the like is used. Then, a top screen for executing various applications installed in the communication terminal 5, a screen related to the executed application (Internet screen, mail screen, navigation screen, etc.), and various information such as images and videos are displayed. In particular, in this embodiment, a map image of an area designated by the user is displayed, and an icon indicating the posted emotion is displayed at the position of the information providing point included in the displayed map image. Further, when the icon displayed on the map image is selected, more detailed information about the information providing point related to the selected icon is displayed.

[0038] In addition, the input operation unit 37 is composed of a touch panel provided on the front surface of the display 36, hard buttons arranged on the housing, and the like. Then, the communication terminal control unit 41 performs control to execute corresponding various operations based on an electrical signal output by pressing the touch panel or the hard button. Note that the input operation unit 37 can also be composed of various keys such as a number / character input key, a cursor key for moving a cursor to select the displayed content, and a determination key for finalizing the selection.

[0039] In addition, the GPS 38 can detect the current position and current time of the communication terminal 5 (that is, the user) by receiving radio waves generated by artificial satellites. In addition to the GPS 38, the communication terminal 5 may be configured to include another device (for example, a gyro sensor, etc.) for detecting the current position and orientation of the communication terminal 5.

[0040] In addition, the transmission / reception circuit unit 39 is a circuit unit for transmitting and receiving signals to and from a base station of the communication network 6 according to a communication standard such as 3G, 4G, or LTE.

[0041] Next, in the information providing system 1 having the above configuration, the sentiment analysis processing program executed by the information providing server 3 will be described with reference to FIG. 6. FIG. 6 is a flowchart of the sentiment analysis processing program according to the present embodiment. Here, the sentiment analysis processing program is executed after a predetermined time (for example, 24 hours) has elapsed since the previous execution, and analyzes the post information stored in the post information DB 13 to determine posts containing negative words and identify the "post sentiment" for each information providing location. Note that the programs shown as flowcharts in FIGS. 6 and 10 below are stored in the RAM 22, ROM 23, etc. provided in the information providing server 3 and are executed by the CPU 21.

[0042] Here, the sentiment analysis processing program executes processing for information providing locations across the country, and the "post sentiment" is identified for each information providing location. However, it is not necessarily required to execute processing for all information providing locations across the country. For example, it is also possible to execute processing only for specific genres or locations of a predetermined scale or larger. Also, the interval for executing the sentiment analysis processing program for each location may be changed.

[0043] First, in step (hereinafter abbreviated as S) 1, the CPU 21 determines whether there is post information associated with the information providing location that is the current evaluation target in the post information DB 13. Note that it is desirable to exclude old post information from the analysis target in order to perform accurate sentiment analysis. Basically, it is determined whether there is post information posted within the last three months or within the last six months. Also, it may be determined whether there is at least one or more post information, but since accurate analysis is difficult with a small number of samples, it may be determined whether there is post information of a predetermined number (for example, 3) or more.

[0044] Here, as shown in FIG. 3, the posted text information DB 13 stores the posted text information obtained in advance from the external SNS server 8. In particular, the content of the posted text (the text data of the message) is stored in association with the location name indicating the location associated with the posted text (the location where the posted text was posted) and the date and time of posting. Therefore, in step S1, it is determined whether there is posted text information associated with the location name of the information providing location to be evaluated.

[0045] Then, when it is determined that there is posted text information associated with the information providing location to be evaluated this time in the posted text information DB 13 (S1: YES), the process proceeds to S2. On the contrary, when it is determined that there is no posted text information associated with the information providing location to be evaluated this time in the posted text information DB 13 (S1: NO), since no posted text to be analyzed exists for the sentiment of the information providing location to be evaluated, the sentiment analysis processing program is terminated.

[0046] In S2, the CPU 21 extracts and acquires the posted text information associated with the information providing location to be evaluated this time from the posted text information DB 13. In addition, in order to perform accurate sentiment analysis, it is desirable to exclude old posted texts from the analysis target, and basically, the posted text information posted within 3 months or within 6 months is acquired as the target.

[0047] Then, the following processes of S3 to S10 are performed for each of the posted text information obtained in S2, and after performing the processes of S3 to S10 for all the obtained posted text information, the process proceeds to S11.

[0048] First, in S3, the CPU 21 performs sentiment analysis on the post information to be processed, thereby predicting the post sentiment, which is the sentiment of the poster who posted the post, especially from the text data indicating the post content. Here, there are various methods for sentiment analysis, such as machine learning, rule-based, and combinations thereof. As an example, in sentiment analysis applying machine learning such as deep learning, by inputting the text data of the post, it is possible to output information identifying the "sentiment category" of the post sentiment from the text data (S4). In sentiment analysis applying machine learning, it is possible to analyze the sentiment from the text data based on elements such as the meaning of words appearing in the sentence, the context of the words used, and the expression method, rather than analyzing individual words by natural language processing technology.

[0049] Particularly in this embodiment, as shown in FIG. 7, the "ratio of sentiment for each sentiment category (sentiment ratio)" is estimated and output as the post sentiment. As shown in FIG. 7, the output sentiment ratio shows the ratio of sentiment for each of the eight basic sentiments of joy, disgust, anger, fear, sadness, expectation, surprise, and trust for each text data to be analyzed. The sum of the ratios of each sentiment is 1, and the closer it is to 1, the stronger the expression of that sentiment in the text data to be analyzed. For example, the sentiment analysis result for text1 indicates that it is a text in which the two sentiments of 'fear' and 'expectation' are strongly expressed respectively. The sentiment analysis result for text2 indicates that it is a text in which only the sentiment of 'expectation' is strongly expressed. The sentiment analysis result for text3 indicates that it is a text in which there is no prominent sentiment and the sentiment is not strongly expressed.

[0050] After that, in S5, the CPU 21 determines whether the posting sentiment in the posting text information to be processed is "negative" based on the analysis result of the sentiment in S4. Specifically, referring to the sentiment ratio (Figure 7) output in S4, if the sentiment category with the highest ratio is any of "fear", "sadness", "disgust", or "anger", which are negative among the eight basic emotions, that is, when it is determined that the posting sentiment belongs to any of "fear", "sadness", "disgust", or "anger", it is determined that the posting sentiment in the posting text information to be processed is "negative". However, in addition to the fact that the sentiment category with the highest ratio is any of "fear", "sadness", "disgust", or "anger", it may be additionally conditioned that the difference between the ratio of the sentiment category with the highest ratio and the ratio of the second-highest sentiment category is equal to or greater than a threshold value (for example, 0.2 or more), that is, the negative sentiment is prominent.

[0051] And when it is determined that the posting sentiment in the posting text information to be processed is "negative" (S5: YES), the process proceeds to S6. On the contrary, when it is determined that the posting sentiment in the posting text information to be processed is not "negative" (S5: NO), it is presumed that the posting text information to be processed does not contain negative words, and the process ends without performing the non-display setting of the posting text. After that, after switching the posting text information to be processed, the processes after S3 are executed again.

[0052] In S6, the CPU 21 determines whether the posting text with the posting sentiment determined to be "negative" contains negative words in particular using the dictionary DB15. Here, words corresponding to negative words are registered in the dictionary DB15 in advance together with the part of speech. The CPU 21, for example, subdivides the text data of the posting text by morphological analysis, discriminates each part of speech, and then determines whether there is a word corresponding to the negative words registered in the dictionary DB15.

[0053] And when it is determined that there is a word corresponding to the negative word registered in the dictionary DB15 (S7: YES), the process proceeds to S8. On the other hand, when it is determined that there is no word corresponding to the negative word registered in the dictionary DB15 (S7: NO), it is presumed that the post text information to be processed does not contain a negative word, and the process ends without performing the non-display setting of the post text (i.e., as a display target). After that, after switching the post text information to be processed, the processes after S3 are executed again.

[0054] Here, even when it is determined that there is a word corresponding to the negative word registered in the dictionary DB15 (S7: YES), it cannot be determined for sure that the post text contains a negative word. In the determination using morphological analysis and a dictionary, it is difficult to make an accurate determination for a text with a complex expression such as a text with a double negative, for example. Even if a word that matches the negative word registered in the dictionary DB15 is included in the text, there are many cases where it is not actually a negative word. In particular, texts posted on SNS are often not in correct grammar, and it is difficult to make an accurate determination for such texts.

[0055] Therefore, in this embodiment, in S8, the CPU 21 uses machine learning to determine again whether the post text information to be processed is a post text containing a negative word. In particular, in this embodiment, machine learning such as deep learning is applied, and a plurality of special emotion categories (hereinafter referred to as NG categories) indicating negative emotions are newly set separately from the above basic emotions, and it is determined whether the post emotion belongs to a plurality of types of NG categories, thereby determining whether it actually contains a negative word. The NG category is an emotion that is expected to be accompanied by a post text that gives discomfort, anger, or sadness when viewed by a third party among negative emotions. In this embodiment, four types, namely, 'disgust and aversion', 'aggressive', 'discrimination','subjectivity (negation based on one's own perspective and emotion)', are set. However, the NG category is not limited to the above four types as long as it is an emotion category indicating a negative emotion, and other emotions can also be included.

[0056] Here, in a storage medium such as the ROM 23 or the flash memory 24 that the information providing server 3 has, a learned learning model 45 used for analyzing whether it belongs to the above NG category is recorded. As the learning model 45, for example, a neural network is used. In particular, as shown in FIG. 8, the learning model 45 of the present embodiment is a learning model that has been previously learned to estimate and output the ratio (analysis ratio) of the posted text belonging to each NG category from the text data of the posted text by inputting the text data of the posted text. Here, in the learning model 45 using a neural network, as shown in FIG. 8, data obtained by multiplying the weight (weight coefficient) by each neuron, which is the output data after performing the processing in the input layer on the text data input to the input layer, is input to the next intermediate layer. Then, similarly in the intermediate layer, data obtained by multiplying the weight (weight coefficient) by each neuron, which is the output data after performing the processing in the intermediate layer, is input to the next output layer. Then, the analysis ratio is finally output from the output layer. Incidentally, in the learning of the learning model 45, as the learning progresses, the above weight (weight coefficient) is appropriately changed to a more suitable value and set.

[0057] Also, in the present embodiment, it is assumed that the information providing server 3 has a learning model 45 that has been sufficiently learned in advance using teacher data or the like, but the learning of the learning model may be performed in parallel with the output of the analysis ratio. It is also possible to learn without a teacher. Furthermore, the learning model 45 does not necessarily have to consist of only one neural network, and may include a plurality of neural networks.

[0058] In addition, as shown in FIG. 8, the analysis ratios output by the learning model 45 predict the ratios belonging to four types of NG categories, namely, "disgust / loathing", "aggressive", "discrimination", and "subjectivity", and the emotion "not bad (Good)" other than these NG categories, for each piece of input text data. The sum of the ratios of each emotion is 1, and the closer it is to 1, the stronger the expression of that emotion in the text data to be analyzed. For example, the emotion analysis results for text1, text3, and text4 indicate that the text is a passage in which negative emotions due to "subjectivity" are strongly expressed. The emotion analysis result for text2 indicates that the text is a passage in which negative emotions due to "disgust / loathing" and "subjectivity" are strongly expressed respectively. The emotion analysis results for text5 and text6 indicate that the text is a passage in which emotions of any of the NG categories are not strongly expressed. In addition, in the analysis results shown in FIG. 8, in addition to whether the posted text contains negative words, it is also possible to grasp more specifically what kind of negative-word-containing posted text it is, that is, the types of negative words.

[0059] In the same way as described above, for each piece of posted text information determined to have a word corresponding to a negative word by morphological analysis and a dictionary, analysis using machine learning is performed (S8), and after outputting the analysis ratio, the process proceeds to S9.

[0060] In S9, the CPU 21 determines again whether the posted text information to be processed contains negative words based on the analysis results using the above machine learning. For example, when the ratio belonging to "not bad" is equal to or less than a threshold value (for example, 0.5), it is determined that the posted text contains negative words. Alternatively, if the criteria for the display target are set loosely, for a text with negative emotions due to "subjectivity", it may be excluded from negative words. In that case, when the sum of the ratios belonging to "subjectivity" and "not bad" is equal to or less than a threshold value (for example, 0.5), it is determined that the posted text contains negative words.

[0061] When it is determined based on the analysis result by machine learning that the post information to be processed is a post containing negative words (S9: YES), the process proceeds to S10. On the other hand, when it is determined based on the analysis result by machine learning that the post information to be processed is not a post containing negative words (S9: NO), it is presumed that the post information to be processed does not contain negative words, and the process ends without performing non-display setting of the post (i.e., as a display target). After that, after switching the post information to be processed, the processes after S3 are executed again.

[0062] In S10, the CPU 21 determines that the post information to be processed contains negative words, and excludes the corresponding post from the display target for the user. Note that the post information DB 13 (Fig. 3) also stores a flag for identifying whether or not to exclude each post from the display target. Specifically, the exclusion setting is performed based on the flag. As described above, in this embodiment, by referring to the analysis ratio for each NG category, it is possible to more specifically grasp what kind of negative words with what kind of emotions are included in the post. In addition to whether or not to exclude the post from the display target, the display target flag also adds information about what kind of emotions ('dislike', 'aggressive', 'discrimination','subjective') the post more specifically has with the negative words it contains, that is, information for identifying the type of negative words. As a result, for example, it is also possible to specify the type of emotion for exclusion setting. It is possible to exclude only the posts with a high analysis ratio of 'dislike', it is possible to exclude only the posts with a high analysis ratio of 'aggressive', and it is possible to exclude only the posts with a high analysis ratio of 'discrimination'.

[0063] Note that the posts excluded from the display target in S10 do not become the display target (browsing target by third parties), but are used as information for specifying the post emotion of the information providing point. Specifically, it is included in the posts aggregated when calculating the emotion value in S11 described later.

[0064] In the same manner as above, for each piece of post information obtained in S2, sentiment analysis (S3) and exclusion setting of display targets using morphological analysis and machine learning are performed (S6 to S10). After executing the process for all the obtained post information, the process proceeds to S11.

[0065] In S11, the CPU 21 aggregates the sentiment ratios output in S4 and calculates a sentiment value, which is the total value of the sentiment ratios for each of the eight types of basic sentiments. The sentiment value for each basic sentiment calculated in S11 is the result showing which basic sentiment has a high sentiment value by classifying and aggregating the results of sentiment analysis of the posts for all the posts submitted for the information-providing location that is the evaluation target this time into multiple types of basic sentiments.

[0066] Next, in S12, the CPU 21 determines the sentiment classification of the information-providing location that is the evaluation target this time based on the tendency indicated by the sentiment value calculated in S11. Here, in this embodiment, sentiment classifications of type 1 to type 3 are defined, and in S12, it is determined which of type 1 to type 3 the sentiment classification of the information-providing location that is the evaluation target this time corresponds to.

[0067] Examples of each of type 1 to type 3 of sentiment classification are shown in FIG. 9. 'Type 1' is a sentiment classification showing a tendency that only one sentiment stands out. As a specific condition, the difference between the sentiment value of the most highly sentimented basic sentiment and the sentiment value of the second most highly sentimented basic sentiment is set to be equal to or greater than the first threshold value. The first threshold value is, for example, 40% of the total value obtained by summing up all the sentiment values of each sentiment. 'Type 2' is a sentiment classification showing a tendency that multiple sentiments stand out. As a specific condition, the difference between the sentiment value of the most highly sentimented basic sentiment and the sentiment value of the second most highly sentimented basic sentiment is less than the first threshold value, and there is at least one basic sentiment whose sentiment value is equal to or greater than the second threshold value. The first threshold value is, for example, 40% of the total value obtained by summing up all the sentiment values of each sentiment, and the second threshold value is, for example, 25% of the total value obtained by summing up all the sentiment values of each sentiment. "Type 3" is an emotion classification that shows a tendency of overall weak emotions without prominent emotions. As a specific condition, it is assumed that there is no basic emotion whose emotion value is equal to or higher than the second threshold. Note that the second threshold is, for example, 25% of the total value obtained by summing up all the emotion values of each emotion.

[0068] After that, in S13, the CPU 21 determines whether the emotion classification of the information providing location to be evaluated this time corresponds to "Type 1". And when it is determined that the emotion classification of the information providing location to be evaluated this time corresponds to "Type 1" (S13: YES), the process proceeds to S14. On the contrary, when it is determined that the emotion classification of the information providing location to be evaluated this time does not correspond to "Type 1" (S13: NO), the process proceeds to S17.

[0069] Subsequently, in S14, the CPU 21 compares the emotion values for each basic emotion calculated in S11 for the information providing location to be evaluated this time, and specifies the basic emotion with the highest emotion value as the posted emotion (the basic emotion to which it belongs) of the information providing location to be evaluated this time. For example, in the example of "Type 1" shown in FIG. 9, "disgust" will be specified as the posted emotion (the basic emotion to which it belongs).

[0070] After that, in S15, the CPU 21 performs morphological analysis on the posted texts associated with the information providing location to be evaluated this time, that is, the posted texts obtained in S2, and keywords indicating the evaluation and situation of the information providing location are extracted from those posted texts. However, the posted texts excluded in S10 are excluded from the target of keyword extraction. Note that a keyword may be set as an extraction target if it is included even once in any of the posted texts, or the condition of being included a predetermined number of times or more (frequent words) may be set as an extraction condition. Also, it is desirable to set an upper limit on the number of keywords to be extracted for information providing locations with a large number of posts. When the upper limit is reached, prioritize and retain the keywords with a large number of extractions (that is, the keywords frequently included in the posted texts).

[0071] Next, in S16, the CPU 21 stores the basic emotion specified in S14 as the "posted emotion" for the information providing point that is the target of the current evaluation in the distribution information DB 7, and also stores the keyword extracted in S15 as the "posted text" posted for the information providing point that is the target of the current evaluation in the distribution information DB 7. As described above, the distribution information DB 7 stores various types of information including information regarding the "posted emotion" and "posted text" for the information providing points that are the targets of information provision across the country (Figure 4).

[0072] On the other hand, in S17, the CPU 21 determines whether the emotion classification of the information providing point that is the target of the current evaluation corresponds to 'Type 2'. And when it is determined that the emotion classification of the information providing point that is the target of the current evaluation corresponds to 'Type 2' (S17: YES), the process proceeds to S18. In contrast, when it is determined that the emotion classification of the information providing point that is the target of the current evaluation does not correspond to 'Type 2', that is, when it is determined that it corresponds to 'Type 3' (S17: NO), the process proceeds to S19.

[0073] Subsequently, in S18, the CPU 21 compares the emotion values for each basic emotion calculated in S11 for the information providing point that is the target of the current evaluation, and specifies the basic emotion whose emotion value is equal to or greater than the second threshold as the basic emotion to which the posted emotion of the information providing point that is the target of the current evaluation belongs. If there are multiple basic emotions whose emotion values are equal to or greater than the second threshold, all corresponding ones are specified as the posted emotions of the information providing point that is the target of the current evaluation. However, it may be up to a predetermined number (for example, the top two) in descending order of emotion value. The second threshold is, for example, 25% of the total value obtained by summing up all the emotion values of each emotion. For example, in the example of 'Type 2' shown in Figure 9, "joy" and "expectation" are specified as the basic emotions to which the posted emotion belongs.

[0074] After that, in S15, the CPU 21 performs morphological analysis on the posted texts associated with the information-providing locations to be evaluated this time, which are the same as in the case of "Type 1", that is, the posted texts obtained in S2, and keywords indicating the evaluation and situation of the information-providing locations are extracted from those posted texts. Further, in S16, the CPU 21 stores the basic emotion specified in S18 as the "posted emotion" for the information-providing location to be evaluated this time in the distribution information DB 7, and stores the keywords extracted in S15 as the "posted texts" posted for the information-providing location to be evaluated this time in the distribution information DB 7.

[0075] On the other hand, in S19 which is executed when it is determined that the emotion classification of the information-providing location to be evaluated this time is "Type 3", the CPU 21 determines that there is no prominent emotion for the information-providing location to be evaluated this time, that is, there is no particularly strong emotion as a whole in the posted text, there is no bias in emotion, and the emotion cannot be specified, so "no emotion (lack of emotion)" is specified as the posted emotion (the basic emotion to which it belongs) for the information-providing location to be evaluated this time.

[0076] After that, in S15, the CPU 21 performs morphological analysis on the posted texts associated with the information-providing locations to be evaluated this time, which are the same as in the case of "Type 1", that is, the posted texts obtained in S2, and keywords indicating the evaluation and situation of the information-providing locations are extracted from those posted texts. Further, in S16, the CPU 21 stores "no emotion" as the "posted emotion" for the information-providing location to be evaluated this time in the distribution information DB 7, and stores the keywords extracted in S15 as the "posted texts" posted for the information-providing location to be evaluated this time in the distribution information DB 7.

[0077] Next, in the information providing system 1, the information providing processing program executed by the information providing server 3 and the communication terminal 5 will be described with reference to FIG. 10. FIG. 10 is a flowchart of the information providing processing program according to the present embodiment. Here, the information providing processing program is a program that is executed after a predetermined application program for obtaining information on the information providing location is started in the communication terminal 5, and provides information on the information providing location to the user. Note that the program shown as a flowchart in FIG. 10 below is stored in the RAM or ROM provided in the information providing server 3 or the communication terminal 5, and is executed by the CPU 21 or the CPU 31.

[0078] First, the information providing processing program executed by the CPU 31 of the communication terminal 5 will be described with reference to FIG. 10. In S21, the CPU 31 starts a predetermined application program (hereinafter referred to as the information providing application) for obtaining information on the information providing location. Note that the information providing application may be a navigation application or a dedicated application program different from the navigation application. It is assumed that the information providing application has been downloaded in advance from a web server or the like and installed in the communication terminal 5.

[0079] Here, when the information providing application is started in the communication terminal 5, first, a map image around the current location is displayed on the display 36 (S22). Note that the map display data for displaying the map image 51 is obtained from the information providing server 3. The map image 51 displayed on the display 36 can be freely scaled or the display target area can be changed based on the user's operation.

[0080] Next, in S23, the CPU 31 transmits a request signal for requesting information on the information providing location included in the map image that is the display target on the display 36 at the current time to the information providing server 3. Note that the request signal includes the terminal ID for identifying the transmitting communication terminal 5 and the location ID for identifying the information providing location included in the map image that is the display target on the display 36 at the current time (the location name or position coordinates may be used instead of the location ID).

[0081] Subsequently, in S24, the CPU 31 receives the information transmitted from the information providing server 3 in response to the request signal transmitted in S23. Note that the information received in S24 is information regarding the information providing point included in the map image to be displayed on the display 36 at the current time, particularly information regarding the type of "posted emotion" specified by the above-described emotion analysis processing program (FIG. 6).

[0082] Subsequently, in S25, the CPU 31 displays an icon indicating the presence of the information providing point at the position where the information providing point exists in the map image around the current position displayed on the display 36. The icon also indicates the type of posted emotion of the information providing point.

[0083] Here, FIG. 11 is a diagram showing an example of the icon displayed in S25. As shown in FIG. 11, a map image 51 is displayed on the display 36, and an icon 52 is further displayed at the position where the information providing point exists in the map image 51. The appearance of the icon 52 mimics a face, and there are three types of expressions, and the difference in expressions indicates the type of posted emotion of the information providing point. Note that in the present embodiment, the "posted emotion" is specified as eight basic emotions in the above-described emotion analysis processing program (FIG. 6), but the icon 52 determines two types, positive emotion (positive emotion) and negative emotion (negative emotion), without distinguishing up to the specific type of emotion, and indicates whether the emotion of the poster is closer to positive emotion or negative emotion, or neutral (neutral) which is neither. However, the type of the icon 52 may be increased to notify up to the specific type of emotion.

[0084] As an example, if the basic emotion specified as the posted emotion includes either "joy" or "trust", it is positive; if it includes any of "aversion", "anger", "fear", or "sadness", it is negative; if it includes either "expectation" or "surprise" or is "unemotional", it is neutral. In addition, when it corresponds to multiple emotion categories (for example, positive and neutral), the emotion category corresponding to the basic emotion with the highest emotional value is selected.

[0085] In addition, if the appearance of the icon 52 is "positive", it has a pleasant expression; if it is "negative", it shows anger; if it is "neutral", it has a blank expression. Therefore, when the user visually recognizes the icon 52, they can easily grasp the posted emotion of the information-providing location. Also, for the icon 52, the display size and display color may be changed according to the magnitude of the posted emotion. For example, the larger the magnitude of the posted emotion, the larger the display size. In addition, for an information-providing location where the posted emotion cannot be determined, it is also acceptable not to display the icon 52, or to display the "neutral" icon 52.

[0086] Also, the icon 52 displayed on the map image 51 is a selection target for the user. In S26, the CPU 31 determines whether it has received an operation in which the user selects any of the icons 52 displayed on the map image 51 based on a signal from the input operation unit 37.

[0087] When it is determined that an operation in which the user selects any of the icons 52 displayed on the map image 51 has been received (S26: YES), the process proceeds to S27. On the other hand, when it is determined that an operation in which the user selects any of the icons 52 displayed on the map image 51 has not been received (S26: NO), the information-providing processing program is terminated.

[0088] Subsequently, in S27, the CPU 31 transmits a request signal to the information providing server 3 to request more detailed information regarding the information providing location corresponding to the icon 52 selected by the user. Note that the request signal includes a terminal ID for identifying the communication terminal 5 of the transmission source and a location ID for identifying the information providing location corresponding to the icon 52 selected by the user (the location name or position coordinates may be used instead of the location ID).

[0089] Thereafter, in S28, the CPU 31 receives the information transmitted from the information providing server 3 in response to the request signal transmitted in S27. Note that the information received in S28 is more detailed information regarding the information providing location corresponding to the icon 52 selected by the user.

[0090] Next, in S29, the CPU 31 displays, on the display 36, more detailed information regarding the information providing location corresponding to the icon 52 selected by the user based on the information received in S28. Note that the information to be displayed includes, in addition to the basic emotions specifically identified as "posting emotions" by the aforementioned sentiment analysis processing program (Fig. 6), also the "posting text" used for the analysis of the posting emotions. However, regarding the "posting text", only the keywords indicating the evaluation and situation of the information providing location are used instead of the full text.

[0091] Here, Fig. 12 shows an example of the information providing screen 53 displayed on the display 36 in S29. In the example shown in Fig. 12, on the information providing screen 53, for example, the name of the information providing location, an exterior photo, information regarding the details of the location such as business hours and contact information, and in addition to the type of basic emotion specifically identified as "posting emotion", the keywords extracted from the "posting text" are displayed.

[0092] Specifically, regarding the "posting emotion", the type of emotion specifically identified from among the eight types of basic emotions in the aforementioned sentiment analysis processing program (Fig. 6) is displayed. Also, when "no emotion (without emotion)" is specifically identified (S19), that fact is displayed. In addition, a list of the extracted keywords is displayed side by side for the "posted text". Note that in the display of the keywords, they may all be displayed in the same size and color of text, or the size and color of the text may be changed for each keyword. For example, for keywords with a large number of extractions (i.e., keywords frequently included in the posted text), it is possible to increase the text size or change the display color to a different one. Also, regarding the arrangement of the keywords, instead of simply displaying them side by side, it is possible to display them such that similar keywords are close to each other, or display important keywords centered, etc.

[0093] Note that the information providing screen 53 shown in FIG. 12 is merely an example, and any display mode may be used as long as the basic emotions specified as "posted emotions" and the "posted text" are displayed in a manner that can be grasped by the user. By the user visually recognizing the information providing screen 53 shown in FIG. 12, the user can obtain an accurate evaluation based on the posted text for the specified location.

[0094] Next, the information providing processing program executed by the CPU 21 of the information providing server 3 will be described. Note that each of the following processes S31 to S36 starts at the timing when the corresponding information from the communication terminal 5 is received. Therefore, the execution order of each step is not necessarily in the order of the smaller step numbers.

[0095] First, in S31, the CPU 21 receives a request signal for information transmitted from the communication terminal 5. Note that the request signal includes a terminal ID for identifying the communication terminal 5 as the transmission source and a location ID for identifying the information providing location corresponding to the icon 52 selected by the user (a location name or position coordinates may be used instead of the location ID).

[0096] After that, in S32, the CPU 21 extracts information regarding the information-providing location requested based on the request signal received in S31. Note that, as described above, the distribution information DB 7 stores various information including information regarding "posted feelings" for information-providing locations targeted for information provision across the country (Fig. 4), but in S32, only the types of "posted feelings" are extracted specifically.

[0097] Subsequently, in S33, the CPU 21 transmits the type of "posted feelings" as information regarding the information-providing location extracted in S32 to the communication terminal 5 that is the transmission source of the request signal received in S31. After that, on the communication terminal 5 that has received the information, the posted feelings of the information-providing location are displayed by the icon 52 on the map image 51 as described above (Fig. 11).

[0098] Next, in S34, the CPU 21 receives a request signal for detailed information transmitted from the communication terminal 5. Note that the request signal includes a terminal ID that identifies the communication terminal 5 that is the transmission source and a location ID (it may also be a location name or position coordinates) that identifies the information-providing location corresponding to the icon 52 selected by the user.

[0099] After that, in S35, the CPU 21 extracts information regarding the information-providing location requested based on the request signal received in S34. Note that, as described above, the distribution information DB 7 stores various information including the "posted text" used for analyzing the posted feelings in addition to the basic feelings identified as "posted feelings" for information-providing locations targeted for information provision across the country (Fig. 4), but in S35, basically all of that information is extracted.

[0100] Subsequently, in S36, the CPU 21 transmits the detailed information regarding the information-providing location extracted in S35 to the communication terminal 5 that is the transmission source of the request signal received in S34. After that, on the communication terminal 5 that has received the information, the information regarding the information-providing location is output as described above (Fig. 12).

[0101] As described in detail above, in the information providing system 1, information providing server 3, and communication terminal 5 according to the present embodiment, a posted text posted on the network is acquired together with location information identifying a location associated with the posted text (S2), and the acquired posted text is analyzed to predict the posting sentiment, which is the sentiment of the poster who posted the posted text (S3). At the same time, the predicted posting sentiment is classified and aggregated for each location associated with the posted text that is the analysis source, and the posting sentiment is specified for each location (S14, S18, S19). Further, while at least a part of the posted text is displayed on the screen in addition to the specified posting sentiment, in the display of the posted text, the posted text including negative words is excluded from display (S10). Therefore, it is possible to exclude from the display target the posted text including negative words that are not desirable for a third party to view. Also, since the posting sentiment is specified using the posted text including negative words that are excluded from the display target, while the posted text including negative words that are not desirable for a third party to view is excluded from the display target, accurate specification of the posting sentiment is possible by considering the posted text including such negative words for the specification of the posting sentiment. Also, after specifying to which of a plurality of types of basic sentiments the posting sentiment of the posted text belongs (S5), it is determined whether the posted text for which the posting sentiment is specified to belong to any of fear, sadness, disgust, or anger among the basic sentiments includes a posted text including negative words (S7, S9), and the posted text determined to include negative words is excluded from display (S10). Therefore, it is possible to exclude from the display target the posted text including negative words, particularly among the posted texts belonging to negative sentiments. In addition to the basic emotions, a plurality of emotional categories indicating negative emotions are set. For a posted text that is identified as belonging to any of fear, sadness, disgust, or anger among the basic emotions, the learning model is further used to analyze whether the posted emotion belongs to a plurality of emotional categories. By doing so, it is determined whether it contains negative words (S8). For a posted text determined to contain negative words, based on the results of the analysis using the learning model, it is further specified which emotional category the posted text containing negative words belongs to (S10). Therefore, not only whether it simply contains negative words, but also specifically what kind of emotions the negative words are accompanied by, that is, the type of negative words, can be specified.

[0102] Note that the present invention is not limited to the above-described embodiment, and it goes without saying that various improvements and modifications can be made without departing from the gist of the present invention. For example, in this embodiment, when displaying the posted text posted at the information providing location as the detailed information of the information providing location, only some keywords of the posted text are extracted and displayed. However, the entire text of the posted text may be displayed. Still, even when the entire text is displayed, the posted text excluded in S10 above is excluded from the display target.

[0103] Also, in this embodiment, when determining whether a posted text with negative emotions actually contains negative words (S5 to S9), first, the determination of negative words is performed using morphological analysis and a dictionary (S6), and then the determination of negative words is performed using machine learning (S8). However, the order may be reversed.

[0104] Also, in this embodiment, the emotional analysis of the posted text performed to identify the posted emotion of the information providing location and the emotional analysis of the posted text performed to determine whether the posted text contains negative words are performed by common processing (S3). However, they may be performed by separate processing.

[0105] In addition, in this embodiment, an example in which the communication terminal 5 is applied to a smartphone has been described. However, if it has a function of outputting information regarding the information providing point, it can also be applied to other types of communication terminals. For example, it can be applied to a mobile phone, a tablet terminal, a personal computer, a navigation device which is an in-vehicle device, etc. Further, when applying it to other than the navigation device, it can also be implemented in a situation where the user moves other than by car, for example, in a situation of moving on foot.

[0106] Also, in this embodiment, the information providing server 3 is configured to perform the sentiment analysis processing program (FIG. 6), but a part of the processing may be executed by the communication terminal 5.

Explanation of Reference Numerals

[0107] 1... Information providing system (posting sentiment prediction system), 2... Information providing center, 3... Information providing server, 4... User, 5... Communication terminal, 6... Communication network, 7... Distribution information DB, 8... SNS server, 9... Storage DB, 11... Server control unit, 13... Posting text information DB, 36... Display, 41... Communication terminal control unit, 51... Map image, 52... Icon, 53... Information providing screen

Claims

1. Post text information acquisition means for acquiring a post text posted on a network together with location information identifying a location associated with the post text; Emotion prediction means for predicting a post emotion, which is the emotion of the poster who posted the post text, by analyzing the post text acquired by the post text information acquisition means; Location emotion identification means for classifying and aggregating the post emotions predicted by the emotion prediction means for each location associated with the post text that is the analysis source, and identifying the post emotion for each location; Post emotion display means for displaying at least a part of the post text associated with the designated location on a screen in addition to the post emotion at the designated location, and having: The post emotion display means excludes and displays a post text including a negative word among the post texts associated with the designated location. An information providing system.

2. The location emotion identification means uses the post text including the negative word excluded from the display target by the post emotion display means to identify the post emotion. The information providing system according to claim 1.

3. Emotion classification means for identifying to which of a plurality of types of basic emotions the post emotion of the post text acquired by the post text information acquisition means belongs; Emotion determination means for determining whether or not a post text identified as belonging to any of fear, sadness, disgust, and anger among the basic emotions corresponds to a post text including the negative word; and having: The post emotion display means excludes and displays a post text determined to include the negative word by the emotion determination means among the post texts associated with the designated location. The information providing system according to claim 1 or claim 2.

4. The emotion determination means: Sets a plurality of types of emotion categories indicating negative emotions separately from the basic emotions; For a post text identified as belonging to any of fear, sadness, disgust, and anger among the basic emotions, further analyzes whether or not the post emotion belongs to the plurality of types of emotion categories using a learning model, thereby determining whether or not it includes the negative word; For a post text determined to include the negative word, it further identifies to which emotion category the post text including the negative word belongs based on the result of the analysis using the learning model. The information providing system according to claim 3.

Images