JP3201355B2 - Sentiment analysis system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an emotion analysis system for estimating the mood of a subject based on the expression of the subject and taking appropriate countermeasures for maintaining mental health.

[0002]

2. Description of the Related Art Hitherto, various programs have been proposed for estimating the mood of a subject from the facial expression of the subject, taking appropriate countermeasures, and restoring the mental health of the subject.
The implementation of this program is done by experienced counselors, and response measures are taken according to their own experience and qualities. Today, there are many operators who need to restore such mental health, and operate a personal computer toward a daily display.

[0003]

However, such conventional counselor mental recovery measures basically include:
Because the counselor's occupation was determining the subject based on the individual's experience and qualities, or the mood at the time of the counseling, differences in the counselor's ability and mental state at that time resulted in variations in the results. Data may lack objectivity and, depending on the person being shy, may be deceiving themselves while facing the counselor, and if the counselor changes, the know-how that the counselor has accumulated until then will be interrupted There was a problem that the reliability of data was reduced.

[0004] Conventionally, since the person himself / herself feels abnormal and consults with an appropriate doctor or counselor, in the case of technostress which originally has poor subjective symptoms, the patient may leave the patient until the problem becomes serious. Yes, the response is delayed, mental recovery may be difficult,
When it comes to counseling, you leave the company early,
There is a problem that physical and economical burdens increase because transportation costs and medical treatment costs are required. Furthermore, in the past, in order to restore the mental health of the subject (patient), the fields that the counselor was good at, such as music, painting, and games, were used. There was a problem that the method was ignored and could not be said to be a method that could flexibly respond to individual subjects.

[0005] The present invention allows objective and highly reliable analysis of the subject's emotions without counseling by a doctor or counselor, and enables appropriate countermeasures to be taken early based on the analysis results. The purpose is to obtain an emotion analysis system.

[0006]

In order to achieve the above object, an emotion analysis system according to the present invention includes a moving image input camera that captures a facial expression of a subject in real time, a facial image and an emotion of the subject. The correlation is accumulated and compared with the information from the moving image input camera to enable mood determination. When there is a problem with the mood determination result, if the mood determination result is not appropriate for the subject, the mood is determined. A mood database for accumulating the judgment results, and further, data of how much the subject was affected by any of the countermeasures is stored in the effect database, and when the mood judgment result is valid Based on the accumulated data, it is possible to judge what kind of countermeasures are effective for the problem information .
After the self-determination of the response measures
To the above effect database,
After a fixed period of time, the collection of the subject's facial expression data is resumed
After this resumption, the effect database
Responds to the problem information based on the result of the effect determination process.
Then, it is possible to determine what kind of response measures are effective .

[0007] Further, the effect determination processing may be performed by using the countermeasure.
Based on the difference between the mood judgment results before and after
Storing data in the effect database.
Is what you do.

[0008] Further, the effect determination processing is performed by the subject.
New data based on the results of self-
Data is stored in the effect database .

[0009]

[0010]

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an emotion analysis system according to the present invention. In FIG. 1, reference numeral 1 denotes a personal computer (hereinafter, referred to as a personal computer) as a computer operated by a subject 2, and 3 denotes an expression of the subject 2 in real time. The moving image input camera 4 to be captured is software installed in the personal computer 1 and has a mood database 4a, an effect database 4b, and a countermeasure 4c. In addition, these mood databases 4
a, the effect database 4b and the countermeasure 4c can be individually updated for each user environment as an independent object. Among them, the countermeasure data 4c can be individually replaced with each data content for each user environment. .

As described above, the personal computer 1 is connected in advance with the software 4 of the system and the moving image input camera 3 linked thereto, and the facial expression of the subject 2 is fetched to the personal computer 1 every moment. Step S1). The software 4 sends the captured expression of the subject to the mood database 4a, and performs a mood determination from this (step S2). As a result of comparison with the mood database (step S3), "OK" is returned. If so, the process proceeds to the next step. On the other hand, if there is no problem, the collection of expression data is continued again. On the other hand, if it is determined that there is a problem, that is, if the specific emotion set in advance in the mood database exceeds a prescribed threshold value, then it is determined whether the system determination is appropriate. Or
The target person makes a judgment (correct judgment) (step S4). If it is determined to be correct, the process proceeds to the next step, and if it is determined to be incorrect, the system stores the result in the mood database 4.
Then, after waiting for 30 minutes (step S5), the collection of facial expression data is restarted.

On the other hand, if it is determined that the measure is correct, an effective measure is selected (step S6), and subsequently the effect database 4b is searched. Is determined from past data (step S7), and the result of this determination is passed to the response measure 4c. Therefore, the response measure 4c outputs a specific response measure based on the passed information. Subsequently, the response measure is autonomously proposed to the target person, and if the proposal is accepted by the target person, the response measure is immediately executed (step S8).
The self-judgment of the effect of the countermeasure is made in a format defined by the effect database 4b (step S9). Then, after making this self-determination, a final effect determination process is requested to the effect database 4b. After the effect is determined, 30
The processing of step S2 and the subsequent steps is executed again with a waiting time of one minute.

On the other hand, in step S8, if the proposal is not accepted by the subject, the result is returned to the effect database 4b, and the proposal is received again. The process from step S2 onward is repeated again with a 30-minute standby time as no action.

Next, the details of the mood database 4a, the effect database 4b, and the countermeasure 4c will be described with reference to FIGS. First, FIG. 3 shows a calculation algorithm of the mood database 4a.
(1-1) fatigue, (1-2) anxiety, (1-3) impatience,
(1-4) Anger, (1-5) distraction, (1-6) depression, (1-7) good mood, (1-8) expressionless parameters are set. Expression data corresponds to these parameters on a one-to-one basis. The “weight” of each parameter is initially all 1. Before shipping as a product, sampling is performed for 100 people in advance, and their average expression is input. Therefore, the moving image data sent from the system is sampled every 5 seconds, which expression of each parameter is closest to which is determined, the closest one is selected (step S21), and this is stored in a buffer. (Step S22). Then, in the inquiry mode (step S23), when three minutes (or six minutes, nine minutes) have elapsed after the start of sampling (step S24), the buffer is referred to at each sampling timing, and which parameter has been determined in the past three minutes. What percentage is occupied, and each parameter is multiplied by a "weight" (step S25).

Subsequently, under the condition that the mode is not the inquiry mode (step S26), it is checked whether or not there is any of the calculated percentages exceeding 80% (step S27). And if it does not exceed 80%
Judge as "No problem" and return the parameter to the system. When the self-determination of the subject is received from the system as “inappropriate”, the weight of the parameter determined to be problematic is multiplied by 0.9 (step S28). When the mode is the inquiry mode (step S26), three minutes (or six minutes)
Minutes) (step S29), this is returned to the system. If it has not elapsed, the average value of the parameter in question is calculated, and the effect database 4b
(Step S30). Then, the weights of all parameters are compared, and the maximum value is optimized to 1. However,
If the minimum value is less than 0.1, it is corrected to 0.1.

FIG. 4 shows a calculation algorithm of the effect database 4b. In this effect database, corresponding measures expected to have an effect on each of the parameters registered in the mood database 4a are registered. In each of the countermeasures, 10 is set as an initial value of the degree of effect. When the effect database 4b receives a request including a problematic parameter from the system, the effect database 4b selects a candidate having the highest degree of effect from the corresponding measure candidates corresponding to the parameter. If there are a plurality of candidates having the same effectiveness, they are selected at random (step S4).
1). The effects database returns the selected candidate to the system. When a review request is received from the system (step S42), the second candidate is selected in the same manner as the first candidate except for the first candidate, and returned to the system. The effectiveness of the rejected candidate is multiplied by 0.9 (step S43).

When a request for re-examination is received, one is randomly selected from all registered countermeasures (including countermeasures already rejected) regardless of parameters, and is returned to the system. 0.9 for rejected candidates
(Steps S44 and S45). When the request for review is received again, "no response" is returned to the system (step S46). On the other hand, after the execution of the countermeasure, the system is requested to determine (subject) self-effect.
In the self-effect judgment, the target person is 1. Effective, 2. Somewhat effective. There is little effect, 4. It is determined whether there is no effect at all (step S47). Effect Database 4
b obtains the self effect judgment and saves the result in a buffer. Also, after obtaining the self-effect determination, the user accesses the mood database 4a at the timing of three minutes, six minutes, and nine minutes,
At each of these timings, the percentage of the parameter in question is checked.

The percentage of the problem parameter before implementation is “A”, 3
The average value of% of times is “B”, and the weight of the self-effect determination result (Effective = 1.2, Slightly present = 1.1, Slightly no = 0.9, No = 0.8) is “C”. ", The new degree of effect is set to (AB) × C, and the final new degree of effect is replaced with the corresponding original degree of effect (step S48). However, if the new degree of effect becomes 0% or less, it is corrected to 1%.

In the response measures, when the system is shipped, the response measures deemed appropriate are already registered in each of the response measures for each parameter. At that time, the subject can re-register, for example, up to three of the five candidates from the bottom to an arbitrary measure. Initial registration measures that have been deregistered are temporarily stored in another part and can be re-registered at any time.

[0021]

As described above, according to the present invention, a moving image input camera that captures the expression of a subject in real time,
The correlation between the expression and the emotion of the subject is accumulated and compared with the information from the video input camera to enable mood judgment. If there is a problem with the mood judgment result, the mood judgment result A mood database that stores the mood determination results when it is not appropriate for the person,
The subject accumulates data on how much mood was affected for which response measure, and when the mood determination result is valid, based on the accumulated data, An effect database that enables the user to determine whether such countermeasures are effective is provided.Especially, by observing subjects who face a device such as a personal computer for a long period of time, it is possible to obtain unconscious expression data. The subject himself / herself can use the self-scoring system to continuously improve the mood database based on the results of the collection and the facial expression judgment of the facial expression, and a push-type structure that works in the form of suggestions from the system to provide the mental health of the subject. The deterioration can be expected to be stopped at an early stage, at which point the subject is given optimal mental health recovery measures in the past Determines from database) you can proposed the subject and also, by the effect subject to self-scoring effect that attained the accuracy of effect database. As a result, the subject's emotions can be objectively and reliably analyzed without counseling by a doctor or a counselor, and an effect is obtained in which appropriate countermeasures can be taken early based on the analysis result. .

[Brief description of the drawings]

FIG. 1 is a block diagram showing an emotion analysis system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an operation of the emotion analysis system shown in FIG.

FIG. 3 is a flowchart showing a calculation algorithm of a mood database in FIG. 1;

FIG. 4 is a flowchart showing a calculation algorithm of an effect database in FIG. 1;

[Explanation of symbols]

 Reference Signs List 1 personal computer (computer) 3 moving image input camera 4 software 4a mood database 4b effect database 4c

Claims (3)

(57) [Claims] 1. A moving image input camera which captures a facial expression of a subject in real time, and a correlation between facial expressions and emotions of the subject is accumulated, and a mood determination is performed by comparing the correlation with information from the moving image input camera. When there is a problem with the mood determination result, if the mood determination result is not appropriate for the subject, a mood database that stores the mood determination result, and how much An effect of accumulating data on the influence and, when the mood determination result is appropriate, enabling to judge what countermeasures are effective for the problem information based on the accumulated data. and a database, following the implementation of the countermeasures, the corresponding measures by the subject
After the self-determination of the
Database, and after a certain period of time
Thus, the collection of the expression data of the subject is resumed, and after this resumption, the effect database stores the effect judgment.
What kind of information does the problem information
An emotion analysis system characterized in that it is possible to determine whether a response measure is effective . 2. The effect determination process according to claim 1 , wherein said effect measures are executed.
New data based on the difference in mood judgment results between before and after the treatment
The emotion analysis system according to claim 1, wherein data is stored in the effect database . 3. The effect judging process is performed by the subject.
New data based on the results of the self-
Emotion analysis system according to claim 1 or 2, characterized in that stored in the effect database.