Sensory evaluation of kokum drinks by fuzzy logic and a simple method

ABSTRACT

Kokum drink formulations were evaluated for their acceptability by fuzzy logic and a new method based on simple calculations. Both the methods showed that the sample containing sugar, cumin, cardamom, and roasted salt at 2, 0.2, 0.2, and 1 g/100 ml juice, respectively, was more acceptable. Quality attributes in general showed under ‘satisfactory’ in both the methods. When similarity values for the quality attributes of kokum samples 1 to 4 were calculated, the results from both the methods were in agreement with each other. The new method is advantageous since it is simple and gave results similar to that of fuzzy logic.

KEYWORDS:

• Kokum
• Fuzzy logic
• Quality attribute
• Sugar
• Similarity value

Introduction

Kokum (Garcinia indica Choisy) is an important fruit, and it is widely consumed in the form of juice, syrup, and squash. Trees of Garcinia indica are indigenous to the western ghats of India, and kokum fruits have culinary and pharmaceutical uses. The ripened kokum fruits are either dark purple or red tinged yellow in colour, and juice preparations are purple or blackish red in colour. It has acceptable flavour and sweet acidic or sour in taste. Kokum fruit has been used as an acidulant in Indian food preparations. Kokum drinks have been consumed for the improvement of digestion and for cooling the body temperatures. Kokum was used in the treatment of dysentery, piles, heart complaints, tumours, acidity problems, and liver disorders.^[1–3]

Kokum contains anthocyanins, kokum butter, hydroxy citric acid, and garcinol. Anthocyanins are known to show antioxidant, anti-inflammatory, and anti-carcinogenic activities. Hydroxy citric acid inhibits fat and cholesterol synthesis; reduces body weight; and lowers lipid accumulation. Kokum butter or oil is a nutritive, softening, astringent, demulcent, and emollient agent. It is used in the preparation of ointments, face creams, and lipsticks. Garcinol shows antioxidant and antimicrobial properties.^[4,5] Kokum drinks gained importance due to their nutritional and health promoting effects. Kokum is most widely marketed as juice. However, there are no reported studies for improving the quality attributes of kokum drinks, and no sensory studies were carried out to evaluate their acceptability. In the present study, an attempt was made to improve the quality attributes of kokum juice. Different formulations were subjected to sensory evaluation; and the data obtained was analysed by fuzzy logic, and another new method for the acceptability of the samples. The new method was based on the simple mathematical calculations.

Materials and methods

Preparation of kokum samples

Fresh and ripened kokum fruits were cleaned, cut into two halves, and seeds were removed. Kokum halves along with sugar (10 g/100 g fruit) were placed in to the transparent glass jars, and jars were kept under sunlight for 8–10 days. Juice was diffused out of fruit due to the concentrated sugar mass, and over a period of time, syrup was formed. This syrup was filtered; and roasted salt (5–8 g/100 ml), cumin seed powder (1 g/100 ml), and cardamom powder (1 g/100 ml) were added to the syrup; and stored at 8°C in the 500 ml glass bottles until further use. Syrup was diluted (1:4) with water;, and 25 ml of the diluted sample was given to the panellists for sensory evaluation. Syrup was allowed to reach room temperature before the dilution. Sample 1 was obtained after diluting the syrup that contains only sugar, and there was no other additive. Sample 2, 3, and 4 were obtained after diluting the syrups containing 8, 6.5, and 5 g of roasted salt per 100 ml syrup, respectively. In addition, these three syrups contain cumin seed powder and cardamom powder (1 g/100 ml).

Selection of panellists and sensory evaluation of kokum samples

Panellists were selected from among the students and staff members of Centre for Emerging Technologies, and School of Engineering and Technology, Jain University, Jakkasandra after screening two times based on 60% success in the triangle test.^[6] One hundred non-smoking and healthy panellists were selected after screening 162 members. They were aged between 19 and 62 yrs; and males and females were 63 and 47, respectively, in number. They were made to understand the different quality attributes chosen for the sensory evaluation, the score chart, and way of scoring.^[7,8] The evaluation was conducted as per the regulations given by the American Society for Testing and Materials.^[8] Four kokum samples were given to the panellists; they were advised to sniff three times in order to judge about the aroma and then to taste and swallow the sample. They were advised to rinse their mouth with water before evaluating the consecutive sample.^[7] Samples were allowed to reach the room temperature before being given for the testing. Panellists were asked to tick against the fuzzy scale factor of their choice for each of the quality attribute of each of the sample after the sensory evaluation. They were also asked to tick their preference of quality attributes of kokum in general. Both the responses were collected on a five point linguistic scale, i.e., ‘not satisfactory’, ‘fair’, ‘medium’, ‘good’, and ‘excellent’ for the kokum samples whereas ‘not at all important’, ‘somewhat important’, ‘important’, ‘highly important’, and ‘extremely important’ for the general quality attributes of the kokum product.

Analysis of sensory data by fuzzy logic

The responses of the panellists were analysed by using fuzzy logic method.^[9–14] The important steps involved were: (i) overall calculation of sensory scores of kokum drinks in the triplets form; (ii) membership function estimation on standard fuzzy scale; (iii) overall membership function computation on standard fuzzy scale; (iv) similarity value estimation and ranking of kokum samples; and (v) ranking of quality attributes of kokum samples in general.

Analysis of sensory data by a method based on simple mathematical calculations

This method of sensory evaluation of kokum drinks involved the following steps which are based on the simple mathematical calculations (SMC): (i) calculation of average of score for each sample under different quality attributes based on the preference of panellists; (ii) calculation of average of scores for quality attributes in general; (iii) calculation of weightage for quality attributes of samples in general; (iv) calculation of overall scores for samples; and (v) finding the acceptability of samples or importance of quality attributes by using the calculated scores and the chart that converts 5-point linguistic scale to 6-point scale.

Average of score for each sample under different quality attributes was calculated based on the preference of the panellists on 5-point sensory scale, and numerical value given to the each factor on 5-point sensory scale (0 = ‘not satisfactory’, 2.5 = ‘fair’, 5 = ‘medium’, 7.5 = ‘good’, and 10 = ‘excellent’). For sample ‘X’ under quality attribute ‘Y’, if ‘b’ number of panellists given ‘not satisfactory, ‘c’ number of panellists given ‘fair’, ‘d’ number of panellists given ‘medium’, ‘e’ number of panellists given ‘good’, and ‘f’ number of panellists given ‘excellent’, then average score for sample ‘X’ under ‘Y’ was calculated as below:

The above equation can be written as below:

(1)

where, S, X, Y, and T represent sample, sample number, quality attribute, and total number of judges, respectively.

For calculating average scores for quality attributes in general, preference of the panellists for each attribute on 5-point sensory scale, and numerical value given to each factor on 5-point sensory scale (0 = ‘not at all important’, 2.5 = ‘somewhat important’, 5 = ‘important’, 7.5 = ‘highly important’, and 10 = ‘extremely important’) were used. For quality attribute ‘Y’, if ‘b’ number of panellists given ‘not at all important’, ‘c’ number of panellists given ‘somewhat important’, ‘d’ number of panellists given ‘important’, ‘e’ number of panellists given ‘highly important’, and ‘f’ number of panellists given ‘extremely important’, then average score for the quality attribute ‘Y’ was calculated by using the following equation;

(2)

where Q, Y, and T represent quality, quality attribute, and total number of judges respectively. Weightage for quality attributes of samples in general was calculated based on the values obtained from Eq. (2) for each quality attribute. If, four quality attributes (‘Y1’, ‘Y2’, ‘Y3’, and ‘Y4’) have been studied, then weightage for the quality attribute ‘Y1’ was calculated by using the following equation

(3)

where Q and W represent quality and weightage of quality attribute, respectively.

Overall scores for samples was calculated by using the average values obtained from the Eq. (1) and the weightage values obtained from the Eq. (3). Overall score for the sample X was calculated by using the following equation:

(4)

where S_O represents overall score of sample.

The scores obtained from the equations 1, 2, and 4 were used to find out the acceptability of the sample or the importance of the quality attribute from Table 1 that converts 5-point linguistic scale to 6-point scale. The scores obtained from Eq. (1), (2), and (4) represent the similarity values for the quality attributes of the given samples, the similarity values for the quality attributes of the samples in general, and the similarity values for the overall ranking of the samples, respectively.

Table 1. Converting 5-point linguistic scale to 6-point scale.

Score (s)	Importance/Acceptability
0–1	Not at all necessary/Not satisfactory
1 < s ≤ 3	Somewhat necessary/Fair
3 < s ≤ 5	Necessary/Satisfactory
5 < s ≤ 7	Important/Good
7 < s ≤ 9	Highly important/Very good
9 < s ≤ 10	Extremely important/Excellent

Results and discussion

The kokum syrup was added with sugar (10%), roasted salt (5–8 %), cumin seed powder (1%), and cardamom powder (1%) to improve the sensory attributes of the kokum drinks. Sample 2, 3, and 4 were obtained after diluting (1: 4 with water) the kokum syrups containing 8, 6.5, and 5% of roasted salt, respectively. These three samples also contain cumin seed powder and cardamom powder at 1% before the dilution. Final concentrations of sugar, cumin, and cardamom powder were 2, 0.2, and 0.2%, respectively, in all the samples whereas roasted salt was at 0, 1.6, 1.3, and 1% in the sample 1, 2, 3, and 4, respectively. Sample 1 obtained after diluting the syrup that contains no additives except sugar. Cumin acts not only as flavouring agent but also as natural preservative due to its antimicrobial property. It is a good source of iron and strengthens the immune system.^[15] It was widely used in the ayurvedic medicine for treating dyspepsia, diarrhoea, and jaundice.^[16] Another additive, cardamom, was widely used in the dishes for its aroma and flavour.^[15] It was used as anti-infective agent and for treating digestive disorders.^[16] Hence, addition of cumin and cardamom to the kokum formulations will improve the aroma and health benefits of the drinks. In addition to the preservation, salt was used in the food materials for improving the taste and texture.^[17] Hence, different concentrations of roasted salt were included in the formulations in addition to cumin and cardamom. Evaluation of the sensory properties of the food samples was commonly practised after changing the ingredients^[18,19] in order to check their acceptability. Here, responses from 100 panellists were obtained and consolidated to get the summary of sensory scores for the quality attributes of kokum samples (Table 2) and that of quality attributes of kokum juice samples in general (Table 3).

Table 2. Summary of the sensory scores for the quality attributes of kokum samples.

Sensory quality attributes and kokum juice samples	Sensory scale factors on 5-point scale
	Not satisfactory	Fair	Medium	Good	Excellent
Aroma
Sample 1	24	21	25	14	16
Sample 2	22	23	28	17	10
Sample 3	28	22	25	18	7
Sample 4	16	20	22	20	22
Colour
Sample 1	20	23	23	17	17
Sample 2	19	20	24	18	19
Sample 3	22	22	26	20	10
Sample 4	15	23	24	24	14
Mouth feel
Sample 1	22	20	26	15	17
Sample 2	20	19	26	19	16
Sample 3	23	22	22	17	16
Sample 4	18	16	27	22	17
Taste
Sample 1	12	20	28	22	18
Sample 2	21	21	30	22	6
Sample 3	19	18	29	20	14
Sample 4	12	21	23	24	20

Table 3. Summary of the sensory scores for the quality attributes of kokum samples in general.

Quality attributes of kokum juice in general	Sensory scale factors on 5-point scale
	Not at all important	Somewhat important	Important	Highly important	Extremely important
Aroma	16	14	26	30	14
Color	16	17	24	28	15
Mouth feel	13	20	24	27	16
Taste	17	15	20	22	26

When similarity values for the overall ranking of the kokum samples were calculated by using fuzzy logic method (Table 4), the similarity values for the sample 1 under ‘not satisfactory’, ‘fair’, ‘satisfactory’, ‘good’, ‘very good’, and ‘excellent’ were 0.04759, 0.34566, 0.708092, 0.653649, 0.264081, and 0.024206, respectively. The highest similarity value, 0.708092, was showed under ‘satisfactory’ category. Similarly, the overall quality of samples 2 and 3 were showed under ‘satisfactory’, whereas for sample 4, it was showed under ‘good’. Among 1, 2, and 3 which were ‘satisfactory’, 3 (0.756863) was better than 2 (0.733362), and 2 was better than 1 (0.708092). After comparing the similarity values of all the samples, the overall ranking is sample 4 (good) > sample 3 (satisfactory) > sample 2 (satisfactory) > sample 1 (satisfactory).

Table 4. Similarity values of the kokum samples for the overall ranking (fuzzy logic). Values in bold indicate the highest similarity value for the respective sample.

Sensory scale factor	Sample 1	Sample 2	Sample 3	Sample 4
Not satisfactory, F1	0.04759	0.062436	0.070662	0.03054
Fair, F2	0.34566	0.386747	0.417814	0.281525
Satisfactory, F3	0.708092	0.733362	0.756863	0.6405
Good, F4	0.653649	0.62689	0.60222	0.685611
Very good, F5	0.264081	0.233447	0.1976	0.34884
Excellent, F6	0.024206	0.017913	0.009782	0.05821

When same data was analysed by using SMC method (Eq. (4)) described in the ‘materials and methods’, the sample 4 was showed under the category ‘good’, whereas all other samples were showed under ‘satisfactory’ (Table 5). These results are similar to those obtained by using fuzzy logic. However, the ranking among the samples that showed under ‘satisfactory’ grade (sample 1 > sample 2 > sample 3) differs from that obtained by using fuzzy logic (sample 3 > sample 2 > sample 1). This should not become important because the difference among the similarity values of the samples under ‘satisfactory’ is not significant in both the methods. In both the methods, the sample 4 containing the aroma improving additives (cumin and cardamom), and roasted salt was accepted over the sample 1 that does not have the above additives. These results suggest that addition of the above additives led to the better acceptability of the kokum drink. Similarly, the addition of vanilla flavour and maltodextrin improved the acceptability of dahi^[20] and mango^[21] drinks, respectively. The fuzzy approach was also successfully applied to the coffee^[22] and tea^[23] samples. When ranking of the quality attributes of kokum juice samples in general was done by adopting fuzzy logic (Table 6), the following order was obtained.

Table 5. Similarity values of the kokum samples for the overall ranking (SMC).

Sample	Sample 1	Sample 2	Sample 3	Sample 4
Scale factor	Satisfactory	Satisfactory	Satisfactory	Good
Score	4.7837	4.563242	4.387864	5.21717

Table 6. Ranking of the quality attributes of the kokum samples in general (fuzzy logic).

Scale factor	Colour	Aroma	Taste	Mouth feel
Not at all necessary F1	0	0	0	0
Somewhat necessary F2	0.0833	0.07619	0.043373	0.07816
Necessary F3	0.7238	0.6952	0.5662	0.6965
Important F4	0.8655	0.8830	0.9099	0.8819
Highly important F5	0.1929	0.2186	0.26212	0.21662
Extremely important F6	0	0	0	0

Taste (important) > Aroma (important) > Mouth feel (important) > Colour (important)

When the ranking was done by using SMC method (Eq. (2)), all the attributes were showed under ‘important’ (Table 7), which is similar to that obtained by using fuzzy logic. However, the order becomes taste > mouth feel > aroma > colour which is slightly different from the one obtained by using fuzzy logic. When all the attributes or samples showed under the same category, the SMC method should not be used for the ranking. However, this method is simple and quickly gives the category under which a given sample or attribute will show.

Table 7. Ranking of the quality attributes of the kokum samples in general (SMC).

Attribute	Colour	Aroma	Taste	Mouth feel
Scale factor	Important	Important	Important	Important
Score	5.225	5.3	5.625	5.325

Similarity values for the quality attributes of kokum samples 1 to 4 were calculated by using fuzzy logic (Table 8) and SMC method (Table 9; Eq. (1)). According to the fuzzy logic with respect to the sample 1, the highest similarity values for ‘color’, ‘aroma’, and ‘mouth feel’ were showed under the category ‘satisfactory’; whereas for ‘taste’ it was showed under ‘good’. Same results were obtained in the SMC method for the sample 1. The results from fuzzy logic, and SMC method were in agreement with each other even in the case of samples 2, and 4. In the case of sample 3, the results from both the methods were in agreement with each other except for the ‘taste’. Here with the minor difference of 0.0043, the highest similarity value (0.6745) for ‘taste’ was showed under ‘good’ instead of ‘satisfactory’ (0.6702) according to the fuzzy logic method. Hence, this result does not contradict much with that of SMC method. There will be no statistical ambiguity in the case of SMC method, since attribute category will be decided based on the one similarity value. However, in fuzzy logic the attribute category will be decided based on more than one similarity value, and the highest one will be taken while categorising the attribute. If there is no significant difference between the similarity values, there will be ambiguity over the statistical significance of the difference between the similarity values in the case of fuzzy logic method. In the SMC method, there is a score range based on which one can estimate how much improvement has to be done to meet the desired category for a sample. However, this is not possible with the fuzzy logic. Finally, both the methods found that sample 4 is better than other samples including the one that does not have any additives (sample 1). Both the methods are equally useful with their own benefits. Fuzzy approach will be easy to adopt when there are less number of samples, whereas SMC method will be advantageous when more number of samples to be analysed. However, while ranking of the samples or attributes that showed under same category, the SMC method should not be used.

Table 8. Similarity values for the quality attributes of the kokum juice samples (fuzzy logic). Values in bold are highest similarity values for the particular attribute of the particular sample.

Sample	Sensory scale	Quality attribute
		Colour	Aroma	Taste	Mouth feel
Sample 1	Not satisfactory	0.061312	0.069166	0.023375	0.0624
	fair	0.3859	0.4206	0.2387	0.3856
	satisfactory	0.7361	0.7623	0.5851	0.73488
	Good	0.6291	0.6000	0.7077	0.6298
	Very good	0.2299	0.190682	0.4195	0.2302
	Excellent	0.01664	0.00794	0.09299	0.0166
Sample 2	Not satisfactory	0.04487	0.08644	0.006625	0.05295
	fair	0.3371	0.4514	0.40074	0.3520
	satisfactory	0.7000	0.7636	0.7433	0.70499
	Good	0.6557	0.5635	0.6159	0.6470
	Very good	0.2717	0.16641	0.2196	0.2647
	Excellent	0.0262	0.0041	0.01492	0.0250
Sample 3	Not satisfactory	0.08339	0.1104	0.03224	0.06778
	fair	0.4450	0.5276	0.2971	0.4039
	satisfactory	0.7627	0.7904	0.6702	0.7473
	Good	0.5717	0.4944	0.6745	0.6160
	Very good	0.1726	0.1121	0.3	0.2118
	Excellent	0.0052	0	0.03232	0.01258
Sample 4	Not satisfactory	0.0510	0.02963	0.020474	0.0364
	fair	0.3372	0.2776	0.2217	0.2992
	satisfactory	0.6854	0.6385	0.5627	0.6569
	Good	0.6499	0.6881	0.7104	0.6684
	Very good	0.2826	0.3535	0.4407	0.3156
	Excellent	0.0299	0.060689	0.1043	0.0433

Table 9. Similarity values for the quality attributes of the kokum juice samples (SMC).

Sample	Colour		Aroma		Taste		Mouth feel
	Scale factor	Score	Scale factor	Score	Scale factor	Score	Scale factor	Score
Sample1	Satisfactory	4.7	Satisfactory	4.42	Good	5.35	Satisfactory	4.62
Sample2	Satisfactory	4.95	Satisfactory	4.25	Satisfactory	4.27	Satisfactory	4.8
Sample3	Satisfactory	4.35	Satisfactory	3.85	Satisfactory	4.8	Satisfactory	4.52
Sample4	Satisfactory	4.97	Good	5.3	Good	5.475	Good	5.1

Conclusion

Kokum drinks were formulated to improve their sensory attributes. Cumin and cardamom were used during the formulation since they were reported to add very good aroma. Roasted salt was used in the formulations since it was shown to improve the taste and texture. Kokum formulations were successfully evaluated by using fuzzy logic and a method based on the simple mathematical calculations. Both the methods showed that the sample containing the cumin and cardamom powders (0.2%) and roasted salt (1%) has better acceptability than the one that does not have the above additives. Sample that has cumin and cardamom will also provide the health benefits due to the medicinal properties of the additives. In both the methods, quality attributes in general were showed under ‘satisfactory’, and similarity values for the quality attributes of all the kokum samples were more or less same. Both the methods were beneficial in their own way with fuzzy logic being easier while handling less number of samples, and SMC method will be advantageous during the analysis of more number of samples.

Acknowledgements

Authors thank Dr. Chenraj Roychand, President, Jain University Trust, and Dr. Krishna Venkatesh, Director and Dean, Centre for Emerging Technologies, Jain University for the research facilities. Authors declare that there is no conflict of interest.