JP2012079144A - Sightseeing spot recommendation device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sightseeing spot recommendation device capable of recommending sightseeing spots suitable for a user's taste.SOLUTION: In the sightseeing spot recommendation device (10), multiple items are preset for each of hierarchy to be displayed on a hierarchy display region 58. When a user selects one of the items, the device displays a question to guide the user to items of other hierarchy related to the selected item on a system prompt display region 70 and waits for the user to select again an item. When the user selects an item of the hierarchy in response to the guiding question, the device calculates spot scores and displays images of multiple sightseeing spots based on the scores on an image display region 60. The sightseeing spots displayed here are recommended sightseeing spots reflecting the user's taste.

Description

  The present invention relates to a sightseeing spot recommendation device and a program, and more particularly, to a novel sightseeing spot recommendation device and a program that can recommend a sightseeing spot that suits a user's preference, for example.

Patent Document 1 discloses a tourist information guide device that displays a tourist guide character string selected by a user and an image relating to the tourist spot on a monitor.
JP 2006-72642 A [G06Q 50/00 G09B 29/00 G09B 29/10]

  Many of the conventional tourist information apparatuses including the above-mentioned background art are ones that a user decides by himself / herself which tourist site he / she wants to know and selects or designates the tourist spot.

  These devices have a problem that the user cannot select an optimum tourist spot unless the user knows “how much the user's own preference matches each tourist spot”. In addition, it is known that the user does not necessarily clarify his / her preferences.

  Therefore, a main object of the present invention is to provide a novel tourist spot recommendation device and program.

  Another object of the present invention is to provide a tourist spot recommendation device and a program that can clarify a user's preference through interaction and can recommend a tourist spot that matches the preference.

  The present invention employs the following configuration in order to solve the above problems. Note that the supplementary explanation and the like here show the correspondence with the embodiments described later in order to help understanding of the present invention, and do not limit the present invention.

  The first invention is an image data storage means for storing image data related to a plurality of tourist spots, and items of each hierarchy related to a plurality of tourist spots obtained by the evaluation grid method are estimated by a method of covariance structure analysis Item storage means for storing together with the degree of relevance between the selected items, question generation means for automatically generating a question related to the item when an arbitrary item of the item storage means is selected according to user input, and questions On the other hand, it is a tourist spot recommendation device comprising image display means for selecting and displaying images of an arbitrary number of tourist spots from the image data storage means based on items answered by the user.

  In the first invention, image data related to each tourist spot is stored in the image storage means corresponding to, for example, the image data area of the embodiment. In the embodiment, the item data storage means corresponding to the item data storage area is stored in the item data storage means. Item data of items to be presented (displayed) to the user is stored. The item data storage means stores each item data of items of each hierarchy related to a plurality of tourist spots obtained by the evaluation grid method together with the degree of relevance between the items estimated by the covariance structure analysis method. In the embodiment, any item in any hierarchy is selected according to the user input, and then the question generating means automatically generates a question related to the selected item according to such user input. Present it to the user. When the user answers the question, that is, when the user selects the next item, the image display means calculates the spot score of each tourist spot from that item, etc. An image is selected from the image data storage means and displayed, and the tourist spot is recommended to the user.

  According to the first invention, when a question generated by the question generating means is further presented to the user in response to the user's item selection, and the next item is selected in response to the question, a tourist spot that meets the user's preference is recommended. To do. Therefore, the user's preference can be clarified through interaction with the user, and tourist spots that match the preference can be recommended.

  A second invention is according to the first invention, further comprising spot score calculation means for calculating a spot score of each tourist spot when the user answers an item to the question, and the image display means is in accordance with the spot score. A sightseeing spot recommendation device that selects an arbitrary number of ranked sightseeing spots.

  In 2nd invention, an image display means displays the image of a sightseeing spot in order in accordance with a user preference according to a spot score.

  A third invention is a tourist spot recommendation device according to the second invention, wherein the spot score is calculated based on a Bayesian probability of each tourist spot for each item.

  In the third invention, calculation is performed based on the Bayesian probability (according to prior probabilities and conditional probabilities) using a known technique such as Naive Bayes or Bayesian network.

  A fourth invention is dependent on any one of the first to third inventions, the item storage means stores a plurality of items for each different hierarchy according to the evaluation grid method, and the question generation means has another hierarchy. This is a sightseeing spot recommendation device that generates a guidance question for guiding to a tourist spot.

  In the fourth aspect of the invention, such a guidance question is generated as a system prompt, and thereby the user's preference can be clarified. Therefore, a tourist spot that accurately reflects the user's preference can be recommended.

  The fifth invention is dependent on any one of the first to fourth inventions, and further comprises an item determining means for determining an item to be displayed in another hierarchy according to the guidance question based on the degree of association. It is a spot recommendation device.

  In the fifth invention, since items to be presented to the user are narrowed down in the next hierarchy, useless items are not displayed, and the user can easily select.

  According to a sixth aspect of the present invention, image data storage means for storing image data related to a plurality of tourist spots, and items of each hierarchy related to a plurality of tourist spots obtained by the evaluation grid method are obtained by a method of covariance structure analysis. A tourist spot recommending program executed by a computer of a tourist spot recommending device, comprising item storage means for storing together with the estimated degree of relevance between the items, the tourist spot recommending program comprising: When an arbitrary item of the storage means is selected, a question generating means for automatically generating a question related to the item, and an image of an arbitrary number of sightseeing spots based on the item answered by the user to the question Sightseeing spot prediction that functions as an image display means to select and display from the image data storage means Is a program.

  In the sixth invention, the same effect as in the first invention can be obtained.

  According to this invention, it is possible to recommend a tourist spot that matches the user's preference.

  The above object, other objects, features and advantages of the present invention will become more apparent from the following detailed description of embodiments with reference to the drawings.

FIG. 1 is an illustrative view showing an appearance of a sightseeing spot recommendation device according to one embodiment of the present invention. FIG. 2 is a block diagram illustrating a configuration of the sightseeing spot recommendation device according to the embodiment. FIG. 3 is an illustrative view showing one example of a RAM memory map of the sightseeing spot recommendation device of the embodiment. FIG. 4 is a table showing the results of a web (web) questionnaire that is the basis of the database acquired in the example. FIG. 5 is a table showing the expected value of each item when a sightseeing spot is given. FIG. 6 is a table illustrating an example of the degree of association between the items in the embodiment calculated by the covariance structure analysis technique. FIG. 7 is a flowchart showing an example of the sightseeing spot recommendation operation in the embodiment. FIG. 8 is an illustrative view illustrating an initial screen in FIG. 7 embodiment. FIG. 9 is an illustrative view showing one example of a search screen in FIG. 7 embodiment. FIG. 10 is an illustrative view showing another example (restricted items) of the search screen in the embodiment of FIG.

  Referring to FIG. 1, a sightseeing spot recommendation device 10 according to an embodiment includes a display device 12 having a relatively large display screen. The display device 12 is displayed on the upper end of a column 16 upright on a base 14. It is attached so that it tilts slightly backward. The display device 12 typically includes a flat display 20 such as a liquid crystal display or a plasma display, which is held by a frame 18, and a screen used by the user is used for an input operation. A touch panel 22 is installed. However, this touch panel 22 functions as a pointing device, and of course, it may be replaced with a pointing device other than the touch panel or voice input.

  The frame 18 of the display device 12 is provided with a microphone 24 for inputting user's uttered voice on the upper side, and with a speaker 26 for outputting voice to the user on the lower side.

  Referring to FIG. 2, the tourist spot recommendation device 10 includes a CPU 28, a ROM 30, a RAM 32, a flash memory 34, a wireless communication circuit 36 and the like in addition to the display device 20, the touch panel 22, the microphone 24 and the speaker 26 described above. . The flash memory 34 stores a database and the like which will be described later. The wireless communication circuit 36 can be used to connect to the Internet or the like as necessary.

  The RAM 32 is provided with a program area 38 and a data storage area 40 as shown in FIG. In the program area 38, various programs preset in the ROM 40 are developed when used. The program area 38 stores necessary programs such as an OS (Operating System), a free description processing program 381, a system prompt selection program 382, a calculation program 383, and a display program 384.

  The free description processing program 381 determines which of the default items preset by the apparatus 10 the preference corresponds to when the user directly writes his / her preference (or input by voice). It is a processing program for specifying. In the embodiment, in order to select a specific item based on the free description, the vector space method is used to calculate the similarity between the preference input in the free description and the item. I try to imitate that I have chosen.

  Therefore, the free description processing program 381 includes a morphological analysis program (for example, MeCab and Chasen: both are trade names) for decomposing free description sentences into morphemes, and a language related to free description sentences and items based on tf-idf. Includes programs that calculate similarity to expressions through vector comparisons. Here, tf is the term frequency in the sentence, and idf is the document frequency. However, the method of extracting keywords according to tf-idf is well known, and the method of calculating the similarity between two multidimensional vectors is also well known, so detailed description thereof is omitted here. .

  The system prompt selection program 382 is a program for selecting a question (hereinafter referred to as “system prompt”) for further deepening the user's preference based on the item selected by the user. In the embodiment, the above-described items are divided into four different levels and presented to the user (displayed on the display device 20).

The first layer of the four layers is “genre”, and includes items describing characteristics (facts) of tourist spots as shown in Table 1, and the second layer is “image”. 2 includes items describing the atmosphere (subjectivity) of a tourist spot, the third level is “activity”, and describes the things you can experience at a tourist spot as shown in Table 3. The fourth hierarchy is “feeling”, and includes items describing emotional benefits obtained when going to a sightseeing spot as shown in Table 4. In the embodiment, the number of items in the first layer is 43 items, the second layer is 41 items, the third layer is 28 items, and the fourth layer is 25 items. The total number is 137 items. Just an example,
[Table 1]
First layer (spot features, genre)

x1, there is a Buddha statue
x2, there is a characteristic building
x3, temples and shrines,
x4, God of marriage and academics is enshrined
x5, world heritage, world heritage, national treasure, important cultural property
x6 ,, rich in nature
x7, .Colored leaves are beautiful
x8, .Sakura is beautiful
x9,, The garden is beautiful
x10,, Good snow scene ...
x39 ,, there are artistic works
x40, .You can see a characteristic permanent exhibition
x41, parks and botanical gardens,
x42 ,. Admission fee is not so high
x43 ,, equipped with toilets and barrier-free facilities [Table 2]
Second level (atmosphere, image)
x44,, small
x45,, spacious
x46 ,, good view
x47,, feeling loneliness
x48,, not very secular
x49,, brilliant
x50,, ornate
x51,, with tradition and history
x52,, fresh impression
x53,, mysterious
x54,, Fantastic…
x80,, Japanese
x81,, impressive
x82,, powerful
x83,, Thrill
x84 ,, Stylish [Table 3]
3rd level (activity, experience)
x85, can walk
x86,, you can eat outside
x87 ,.
x88,, can relax
x89,, good photo
x90, touching the lives of people in Kyoto
x91,, learning history
x92, .You can see and experience the real thing
x93 ,. New discovery
x94, become knowledge and experience
x95,, Looks good for beauty and health ...
x108, .Look back at yourself
x109, .I can think of many things
x110 ,.
x111, ... you can go ask
x112, Easy to go [Table 4]
4th layer (feeling, mood)
x113 ,, calm down
x114,, relaxing
x115 ,, you can feel calm
x116,, relieved
x117 ,.
x118,, can touch
x119,, warming
x120,, fun
x121,, fresh
x122,, heart is clear
x123, ... I feel great ...
x133 ,, free from stress
x134 ,, motivated to work hard
x135 ,.
x136, I think Japan is good
x137,, You can feel fresh The hierarchical structure as described above is well known as “Evaluation Grid Method” (trade name: Junichiro Sakurai). What people perceive and how they perceive In this example, the evaluation of each tourist spot is expressed in a hierarchical structure.

  In the sightseeing spot recommendation device according to the embodiment, a method called “laddering” is used to ask a question in order to extract a subordinate concept (ladder down) or a superordinate concept (ladder up) for any item answered by the user. This question is registered in advance as a system prompt data 422 in a data storage area 40 (FIG. 3) described later.

  For example, if the user selects the ninth item “Garden is beautiful” from the first layer, the sightseeing spot recommendation device is a system such as “What is the garden beautiful? A prompt and a narrowed list of second-level items are displayed and prompted to select from them. Further, after an item is selected from the second layer, the third and fourth layers are changed while further narrowing down the questions. That is, when starting from the first hierarchy, the transition is “1 → 2 → 3 → 4”. There are four types of transitions defined: “1 → 2 → 3 → 4”, “2 → 1 → 3 → 4”, “3 → 2 → 1 → 4”, and “4 → 3 → 2 → 1”. The transition is not limited to these, and other transitions such as “4 → 1 → 2 → 1 → 4 → 2 → 1” are also conceivable.

  The system prompt selection program 382 is a program that selects an optimal system prompt to bring out the user's further preference based on the item selected by the user, and shares the function of the question generation means.

  Returning to FIG. 3, the calculation program 383 is a program for calculating the spot score and the next layer item relevance each time the user selects an item. However, when the item relevance is calculated in advance and registered in the database together with each item, this calculation program 383 does not need to calculate each time.

  Further, the spot score is a Bayesian probability expressed by, for example, Equation 1, indicating the probability that each tourist spot is selected for the item based on the item selected or designated by the user.

However, P (Sj) is the prior probability that the spot Sj is selected, and P (Sj | x1, ..., xm) should be calculated as "the tourist spot is selected when the user selects the item x1, ..., xm. Posterior probability ". P (x1,..., Xm | Sj) is a conditional probability. In order to obtain P (Sj | x1,..., Xm), a known technique such as naive Bayes or Bayesian network is used.

  Here, in order to deepen the understanding of the above probabilities, a database created by the inventors in the present embodiment will be described.

  The inventors first conducted an “evaluation grid method” interview with residents in Kansai, and prepared a draft of the item group. In addition, we conducted a web questionnaire (hereinafter, questionnaire 1) to about 1000 people, asked questions about why and why about sightseeing spots in Kyoto, and clustered the resulting language expressions from the perspective of synonyms, A total of 137 items were confirmed.

  Next, a web questionnaire (hereinafter, questionnaire 2) is performed again, and for each sightseeing spot that each person wants to visit, a score ("1"-"7") is returned for each item, and the score is set to a threshold value, for example, "3" .5 "as a boundary. For example, if the score for an item is 2, it is “0” because it is 3.5 or less. If the score is 4, it is “1”. The result is illustrated in FIG.

  In Questionnaire 2, a minimum of 30 respondents were collected for each tourist spot, and a total of 2,000 respondents received responses. Answers from respondents at each tourist spot were averaged to obtain the expected value of the conditional probability for each item. FIG. 5 shows the expected value of the conditional probability of each item for a certain tourist spot.

  The above-mentioned prior probability represents how many of the total respondents want to go to a certain sightseeing spot. For example, when 100 out of 2000 respondents want to go, “0.05” is set.

  The next-layer item relevance is a degree indicating the strength of the causal relationship between the items shown in Table 1-4. The purpose of the degree of association is used to generate a question regarding an item in another hierarchy related to the item when an item in a certain hierarchy is selected. In other words, since several tens of items are included in each hierarchy, the item list is narrowed down according to the degree of relevance of the next hierarchy item, so that unrelated items are not presented and the convenience of the user is achieved.

  The database of the embodiment is stored in the flash memory 34 shown in FIG. 2 together with the “raw data” of the previous questionnaire. Based on the result of the questionnaire, each of the databases is applied by applying a covariance structure analysis method. A database is constructed by estimating the coupling coefficient between the items of the hierarchy (numerical value indicating the degree of relevance between the corresponding items: the above-mentioned “relationship degree of the next layer item”).

  FIG. 6 exemplifies the degree of association by such a covariance structure analysis. In FIG. 6, for example, the first item of “genre” that is the first layer is “0.7”, “4” for the first layer, the second item, and the fourth item of the upper layer, that is, the second layer, respectively. It has coupling coefficients (relevance) of “0.3” and “0.5”. For example, the first item of “image” that is the second layer has a degree of association of “0.2” and “0.1” with respect to the first item and the fourth item of the third layer, respectively. However, in FIG. 6, the degree of association other than the adjacent hierarchy is not illustrated, but the degree of association is also calculated for the jump hierarchy sandwiching the adjacent hierarchy.

  As described above, when narrowing down the list of items that can be selected by the user in the next hierarchy along with the system prompt for the item selection in a certain hierarchy of the user, the coefficient (item relevance) by the covariance structure analysis here is used. Choose a bigger one.

  Returning to FIG. 3 again, the display program 384 is a program for displaying the items for each hierarchy as described above, and further displaying the system prompt selected for it. As described above, each time the user selects an item, a spot score is calculated. An image of a tourist spot with a higher rank of the spot score (which may be a still image or a moving image) is also displayed by this display program 384. Displayed.

  The data storage area 40 of the RAM 32 stores the item data 421 including the item names and language expressions of the above-described layers, system prompt data 422, and sightseeing spot image data (moving image or still image) 423. Therefore, the display program 384 displays items, system prompts, and images using each data in the data storage area 40.

  Note that a GUI screen in FIG. 8 and later to be described later is also displayed in accordance with image data set in advance in the image data 423.

  FIG. 7 is a flowchart showing the operation of the sightseeing spot recommendation device 10 of the embodiment. In the first step S1, the CPU 28 (FIG. 2) uses the display program 424 (FIG. 3) to display the initial screen 50 shown in FIG. Is displayed.

  In this initial screen 50, the main menu selected by the main menu tab 52 is displayed. In the main menu, for example, a name input area 54 for inputting the user's name, and the user can freely write a request (preference). A description area 56 and a hierarchy selection area 58 for first selecting one of the four hierarchies described above are formed.

  An image display area 60, an item display area 62, and a redo key 64 are formed on the right side of these areas 54-58. The image display area 60 is a place for displaying an image and a spot score of a tourist spot ranked higher in accordance with a spot score calculated each time the user selects an item. The item display area 62 is an area for displaying which item is selected by the user at that time in the search mode opened by the search tab 66.

  Such an initial screen 50 is displayed in step S1, and in subsequent step S3, the CPU 28 determines whether or not the user has selected free description input.

  When it is determined that the user has selected the free description input, “YES” is determined in the step S3, and the CPU 28 receives the input in a succeeding step S5. When it is detected in step S7 that the description input has been completed under the condition that there is no description input for a certain period of time, for example, in the next step S9, the similarity is calculated by the vector space method described above and written by the user. It is determined which of the 137 items shown in Table 1-4 above corresponds to the user's preference. According to the determination result, in step S11, the CPU 28 shifts to the search mode (FIGS. 9 and 10) and selects an item.

  On the other hand, if it is not a free description, “NO” is determined in step S 3, and the CPU 28 determines whether or not the user has selected any hierarchy in the hierarchy selection area 58 in the next step S 13 based on the touch position of the touch panel. To do. Since it is not a free description, the process waits until the hierarchy selection is confirmed in step S13. If the user does not input the hierarchy selection indefinitely, a message requesting again whether it is free description or hierarchy selection may be displayed.

  If “YES” can be confirmed in step S13, in the next step S15, the CPU 28 shifts to the search mode (FIGS. 9 and 10) and displays the item list. Then, when the user selects any item in step S17, the search screen 68 (FIGS. 8 and 9) is displayed in the next step S19, similarly to the case where the item is selected in the previous step S11.

  In this search screen, items of the hierarchy selected in step S11 or step S13 are displayed in the item display area 72, a system prompt is displayed in the system prompt area 70, and further, an image of a tourist spot based on the spot score is displayed. It is displayed in the image display area 60. That is, the tourist spot related to the image displayed in the image display area 60 is a tourist spot recommended to the user by the tourist spot recommendation device 10 of this embodiment.

  In next step S <b> 21, the CPU 28 determines whether or not the ending key 74 formed at the lower part of the search screen 68 is touched. When the ending key 74 is touched, “YES” is determined in the step S21, and the sightseeing spot recommendation operation in this embodiment ends.

  In step S23, the CPU 28 determines whether the return key 76 or the next key 78 is touched. When either key 76 or 78 is touched, the process proceeds to step S27. In step S27, the system prompt of the next hierarchy is selected according to the system prompt selection program and the relevance of the next hierarchy item is calculated. However, as described above, when the item relevance is calculated in advance and registered in the database together with each item, it is not necessary to calculate in step S27. Then, as shown in FIG. 10, a search screen 68 including the number of items in the next layer narrowed down based on the item relevance is displayed in step S19. At the same time, the spot score is calculated in step S27, and the image of the tourist spot selected in the order according to it is displayed in the image display area 60 as the recommended tourist spot in step S19.

  The automatic selection of the next level system prompt in step S27 corresponds to the question generation means. This is because, as described above, the system prompt is an additional question or a guidance question related to the item selected according to the user's input, and the question is automatically determined in step S27. .

  When “NO” is determined in step S23, in step S25, the CPU 28 waits for the user to select one of the question items displayed at that time. That is, if the return key 76 or the next key 78 is not touched in step S23, the user intends to select an item on the search screen 68 at that time, and therefore waits until the user selects an item in step S25. If the user does not input item selection indefinitely, a message requesting item selection may be displayed. When the user selects any item in step S25, the process proceeds to step S27, and the operation in step S27 as described above is executed.

  Such a series of operations shown in steps S19 to S27 is repeatedly executed until the operation of the key 74 ending in step S21 is confirmed or an arbitrary fixed number of times.

  In the above-described embodiment, the description has been made on the basis of handwritten input on the touch panel 22 as a method of free description by the user. However, voice data may be input using the microphone 24 (FIG. 1).

  Further, in the above-described embodiment, the description has been made so as to recommend sightseeing spots in the area of Kyoto. However, the area is not limited, and it is easy to change to recommend other national or foreign tourist spots even in other areas.

DESCRIPTION OF SYMBOLS 10 ... Tourist spot recommendation apparatus 20 ... Display device 22 ... Touch panel 28 ... CPU (computer)
32 ... RAM
36 ... Program area 40 ... Data storage area

Claims (6)

Image data storage means for storing image data related to a plurality of tourist spots,
Item storage means for storing items of each hierarchy related to a plurality of tourist spots obtained by the evaluation grid method, together with the degree of association between items estimated by the method of covariance structure analysis,
When an arbitrary item in the item storage means is selected according to the user's input, a question generating means for automatically generating a question related to the item, and an arbitrary based on the item answered by the user to the question A tourist spot recommendation device comprising image display means for selecting and displaying images of a number of tourist spots from the image data storage means. When the user answers an item to the question, further comprising a spot score calculating means for calculating a spot score of each tourist spot,
The tourist spot recommendation device according to claim 1, wherein the image display means selects an arbitrary number of tourist spots ranked according to a spot score.   The tourist spot recommendation device according to claim 2, wherein the spot score is calculated based on a Bayes probability of each tourist spot for each item. The item storage means stores a plurality of items for each different hierarchy according to the evaluation grid method,
The sightseeing spot recommendation device according to any one of claims 1 to 3, wherein the question generation means generates a guidance question for guiding to another level.   The sightseeing spot recommendation device according to claim 4, further comprising: an item determination unit that determines an item to be displayed in another hierarchy according to the guidance question based on the degree of association. Image data storage means for storing image data related to a plurality of tourist spots, and relations between items estimated by the method of covariance structure analysis for items of each hierarchy related to a plurality of tourist spots obtained by the evaluation grid method A tourist spot recommending program executed by a computer of a tourist spot recommending device, comprising item storage means for storing, the tourist spot recommending program comprising:
When an arbitrary item in the item storage means is selected according to the user's input, a question generating means for automatically generating a question related to the item, and an arbitrary based on the item answered by the user to the question A tourist spot recommendation program that functions as image display means for selecting and displaying images of a number of tourist spots from the image data storage means.

Images