WO2021107735A2 - Method for classifying coffee raw materials, method for providing roasting profile of raw beans using same, and system for providing information on coffee raw materials - Google Patents

Abstract

The present application relates to a method for classifying coffee raw materials in a rapid and objective manner, a method for providing a roasting profile of raw beans using same, and a method for predicting the flavor of coffee raw materials. The present application also relates to a system for providing coffee information capable of providing a variety of information on coffee raw materials, established by an objective method.

Description

A method for classifying coffee raw materials, a method for providing a roasting profile of green beans using the same, and a system for providing information on coffee raw materials

This application claims the benefit of priority based on the Republic of Korea Patent Application No. 10-2019-0156268 filed on November 29, 2019, and all contents disclosed in the documents of the Korean patent application are incorporated as a part of this specification.

The present application relates to a method for classifying coffee raw materials, a method for providing a roasting profile of green beans using the same, and a method for predicting the flavor of coffee raw materials.

The present application also relates to a system and method for providing information on coffee raw materials.

In general, coffee raw materials may be divided into coffee cherries, green coffee beans, and coffee beans. Coffee cherries are the fruit of a coffee tree belonging to the genus Coffea of the family Familia Rubiaceae. After harvesting (picking) coffee cherries, the pulp (pulp) is peeled off through processing, and the dried parchment state is called coffee green bean, and coffee beans roasted by applying heat to the coffee beans ( Coffee beans).

Worldwide, the demand and variety for high-quality coffee (specialty) is growing rapidly. However, in many coffee farmers and industries, the prediction of coffee quality and the evaluation of actual taste depend only on the subjective opinion of experts. This subjective method may have different results depending on the expert, and there is a problem in that the results vary depending on time and environment even for the same person. Therefore, the method according to the subjective opinion of experts is not suitable for maintaining or managing the taste of various high-quality coffees.

Therefore, it is a method of classifying coffee raw materials in an objective method, using the method to provide a roasting profile of green coffee beans, a method for rapidly and objectively predicting the flavor of coffee beans at the level of green coffee beans, and providing information formed by this method system is required.

An object of the present application is to classify coffee raw materials in a rapid and objective way, and to provide a roasting profile method of green coffee beans and a method of predicting the flavor of coffee raw materials according to the classification. The present application also aims to provide a coffee information providing system capable of providing various information on coffee raw materials formed by an objective method.

A method for classifying coffee raw materials according to an embodiment of the present application includes providing analysis data by analyzing chemical characteristics of green coffee beans or coffee beans; providing reduced data by reducing the dimension of the analysis data; and analyzing the reduced data to form a cluster.

The method of analyzing the green coffee beans or the chemical properties of coffee beans is characterized in that it is a spectroscopic analysis method.

The step of providing the reduced data is characterized by using a linear dimensionality reduction method (Linear Dimensionality Reduction Methods) or a non-linear dimensionality reduction method (Non-Linear Dimensionality Reduction Methods).

The step of providing the reduced data may include Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Factor Analysis, Singular Vector Decomposition (SVD), or t- It is characterized by using t-distributed Stochastic Neighbor Embedding (t-SNE).

As an example, the forming of the cluster is characterized by using a supervised learning method or an unsupervised learning method.

The step of forming the cluster is characterized by using partitioning, K-means, K-representative object (K-medoid), CLARA (Clustering Large Applications) or CLARANS method.

The method of classifying coffee raw materials of the present application further comprises predicting a cluster through machine learning based on green coffee beans or chemical characteristic analysis data of coffee beans, and dimensionality reduction data of the analysis data. characterized.

The machine learning is a k-nearest neighbor algorithm, logistic regression, naive Bayes classification, stochastic gradient descent, decision tree ), Random Forest, AdaBoost, Gradient Boosting, Support Vector Machine, or Linear discriminant analysis (LDA).

According to another embodiment of the present application, a method for providing a roasting profile of green coffee beans includes: forming a cluster of green coffee beans according to the classification method; and providing a roasting profile for the formed cluster.

Providing a roasting profile for the clusters may include: measuring a chemical content of the clusters formed; and designing a roasting profile of green coffee beans based on the measured chemical composition.

The chemical components include monosaccharides, polysaccharides, lipids, organic acids, proteins and water.

A method for predicting the flavor of coffee raw materials according to another embodiment of the present application includes forming a cluster of coffee raw materials according to the classification method; measuring the chemical component content of the formed cluster; measuring the content of chemical components in the coffee raw material sample; and comparing the content of the chemical component of the coffee raw material sample with the content of the chemical component of the cluster.

Coffee information providing system according to another embodiment of the present application includes a coffee information management server and at least one terminal capable of accessing the coffee information management server through a network.

The coffee information management server receives a request for information on coffee raw materials from a terminal and a transceiver for transmitting and receiving signals or information, a database for storing coffee information, and the terminal connected through a network, and a control unit equipped with a classification information providing module of coffee raw materials for controlling the transceiver to extract information about the raw material from the database and transmit it to the terminal, and the information on the coffee raw material is classified by analyzing the chemical characteristics of the coffee raw material.

In the coffee information providing system, the chemical properties of the coffee raw materials include the types and contents of chemical elements, chemical functional groups, or chemical components included in the coffee raw materials.

The control unit may further include a roasting profile information providing module for receiving a request for information on coffee raw materials from a terminal connected through a network, and controlling the transceiver to extract and transmit the requested roasting profile information of the coffee raw material to the terminal. can

The roasting profile information is characterized in that it is formed based on the classification information of the coffee raw material.

The control unit receives a request for information on coffee raw materials from a terminal connected through a network, extracts the requested characteristic information of coffee raw materials, and further includes a coffee raw material characteristic information providing module for controlling the transceiver to transmit to the terminal can do.

The characteristic information of the coffee raw material may be the type and content of a chemical component included in the coffee raw material. In addition, it is characterized in that the characteristic information of the coffee raw material is provided simultaneously with the requested coffee raw material and the characteristic information of the different coffee raw material.

The control unit receives a request for information on coffee raw materials from a terminal connected through a network, extracts the requested flavor prediction information of coffee raw materials, and further includes a coffee raw material flavor information providing module that controls the transceiver to transmit to the terminal. may include

The control unit may further include a user management module for receiving and managing login information of a user from a terminal accessed through a network.

The control unit may further include a purchase management module for controlling the transceiver to receive payment information of coffee raw materials from a terminal connected through a network, perform a requested payment process, and transmit the result to the terminal.

The method according to an example of the present application can classify coffee raw materials in a quick and objective way, provide a roasting profile of green coffee beans according to the classification, and can quickly and objectively predict the flavor of coffee beans. The method according to another example of the present application may provide various information on coffee raw materials formed by an objective method.

1 is an exemplary view showing the results of principal component analysis on 54 green bean samples in the method of classifying coffee raw materials according to an embodiment of the present application.

2 is an exemplary view showing the results of forming clusters for 54 green bean samples in the method of classifying coffee raw materials according to an embodiment of the present application.

3 to 9 are exemplary views showing normalized average values of chemical component contents for each cluster formed by the method of classifying coffee raw materials according to an embodiment of the present application, respectively.

10 is a block diagram showing the configuration of a coffee information providing system according to an embodiment of the present application.

11 is a block diagram showing the configuration of a terminal of the coffee information providing system according to an embodiment of the present application.

Figure 12 is a block diagram showing the configuration of the coffee information management server of the coffee information providing system according to an embodiment of the present application.

Hereinafter, the present application will be described in detail through Examples, but the scope of the present application is not limited by the Examples below.

A method of classifying coffee raw materials according to an embodiment of the present application includes providing analysis data by analyzing chemical characteristics of green coffee beans or coffee beans; providing dimension-reduced data by reducing the dimension of the analysis data; and analyzing the reduced data to form a cluster.

Currently, a common method of classifying coffee raw materials such as green coffee beans or coffee beans is to consider the coffee variety, size of green coffee beans, production area, producer or growing environment, and the like. However, the classification method of the coffee raw material of the present invention can provide an objective classification criterion than the classification according to the general method because it is classified based on the analysis data on the chemical properties of the coffee raw material.

As an example, the chemical properties of the green coffee beans or coffee beans include the type and content of elements or functional groups included in the green coffee beans or coffee beans. In addition, the green coffee beans or the chemical properties of coffee beans may further include the type and content of chemical components contained in the green coffee beans or coffee beans.

As an example, the chemical component may include carbohydrates (monosaccharides, disaccharides or polysaccharides), organic acids, chlorogenic acids, nitrogenous compounds (proteins, amino acids, trigonelline or caffeine), lipids (triglycerides, fatty acids or diterpenes) or water. .

A method of analyzing such chemical properties may use a spectroscopic analysis method. The spectrochemical method is not particularly limited as long as it can analyze the chemical properties of green coffee beans or coffee beans, and for example, nuclear magnetic resonance (NMR), infrared spectroscopy (IR), Fourier transform infrared spectroscopy (FTIR Spectroscopy), Near-infrared spectroscopy (NIRs) or attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR Spectroscopy) may be used. Preferably, it is possible to use near-infrared spectroscopy (NIRs), which can obtain analytical data on the chemical properties of coffee raw materials non-destructively and quickly without undergoing a pre-treatment process. The near-infrared wavelength region is in the range of 800 nm to 2,500 nm, and the near-infrared spectroscopy method absorbs light at a specific wavelength among the wavelength region of the above range depending on the functional group, and through this, the chemical properties contained in the coffee raw material analysis data can be obtained.

The amount of data that can be obtained is limited because the conventional method for analyzing the chemical properties of green coffee beans or coffee beans measured only the content of specific individual components (values of only a few peaks among many peaks). For example, if you measure the content of 10 individual chemical components in one green coffee bean sample, only 10 data can be obtained from one sample. However, in the present application, chemical properties such as a wide variety of elements or functional groups included in the sample are analyzed rather than individual chemical components present in the sample. Therefore, the number of (all peak values obtainable in an arbitrary wavelength region) that can be obtained from one green coffee bean sample is about 1500 or more. For example, when measuring the content of individual components in the related art, 3,000 data can be obtained from 300 samples, but when measuring properties related to elements or functional groups, more than 450,000 data can be obtained from the same number of samples. to be.

The step of reducing the dimension of the present application may use a linear dimensionality reduction method (Linear Dimensionality Reduction Methods: LDRM) or a non-linear dimensionality reduction method (NLDRM).

In one embodiment, the step of reducing the dimension may include Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Factor Analysis, Singular Vector Decomposition (SVD) or t-distributed Stochastic Neighbor Embedding (t-SNE) may be used, but is not limited thereto.

Dimension reduction refers to a method of reducing the features of analysis data according to the purpose. As for the analysis data on the chemical properties of coffee raw materials measured by spectroscopic analysis on green coffee beans or coffee beans, the characteristics of the analysis data increase as the number of samples to be analyzed increases, and the dimension of the analysis data increases accordingly. Therefore, dimension reduction is required for faster and more efficient data analysis. As for the method of reducing the dimension of the analysis data, an appropriate method of reducing the dimension may be selected according to a feature of the analysis data.

As an example, the method of classifying coffee raw materials may further include processing the analysis data before reducing the dimension of the analysis data. For example, in order to reduce a variable normalized according to the color of the coffee raw material, a processing step of excluding the analysis data measured in the wavelength range of the visible ray band may be included. Through data processing, more meaningful analysis data of the chemical properties of coffee raw materials can be obtained.

The step of forming the cluster of the present application is performed by cluster analysis. Cluster analysis is a data mining method that defines a data group (cluster) considering the characteristics of given data and finds a representative point that can represent the data group. In the present application, a cluster means a group of chemical property data of coffee raw materials having similar properties.

As an example, forming the cluster may use a supervised learning method or an unsupervised learning method. The supervised learning method is a method using an answer label (answer), and the unsupervised learning method is a method not using an answer label (answer). The supervised learning forms a cluster of reduced data by finding correlations or attributes of chemical characterization data of reduced-dimensional coffee raw materials.

On the other hand, the unsupervised learning method can form a cluster only with the analysis data of the chemical component of the coffee raw material in which the correct answer label is unnecessary and the dimension is reduced. Therefore, when there is no correct answer label, it may be more appropriate to form a cluster of coffee raw materials to be analyzed using an unsupervised learning method.

As an example, the step of forming the cluster may use partitioning, K-means, K-representative object (K-medoid), CLARA (Clustering Large Applications), or CLARANS method. An appropriate cluster method may be used in consideration of the convenience of cluster formation, the ease of verification, and the reliability of the cluster. A more meaningful cluster can be formed by analyzing the reduced data using the cluster method as described above.

As an example, forming the clusters may include finding the number of clusters to be formed. For example, an appropriate number of clusters formed from the reduced analysis data may be found through an Elbow, a silhouette, or a gap statistic method. When a cluster of reduced analysis data is formed using the derived number of clusters, it is more efficient to secure a meaningful cluster.

In addition, the step of forming the cluster may further include the step of reclassifying the formed cluster to form a sub-cluster. In the step of forming the sub-cluster, the method used in the step of forming the cluster may be used, and a description thereof will be omitted.

As an example, the method for classifying coffee raw materials may include predicting clusters through machine learning based on green coffee beans or chemical characteristic analysis data of coffee beans, and dimensionally reduced data of the analysis data.

In one embodiment, a cluster of unknown green beans or coffee beans can be easily predicted by machine learning the unknown green beans or chemical characterization data of coffee beans based on data performed in each step. Therefore, it is possible to easily predict a cluster based on the chemical characteristics analysis data of green coffee beans even for new green coffee bean varieties or green coffee beans with a changed cultivation environment.

As an example, the machine learning is K-nearest neighbor algorithm, logistic regression (Logistic Regression), naive Bayes classification (Naive Bayes Classification), stochastic gradient descent (Stochastic Gradient Descent), Decision Tree, Random Forest, AdaBoost, Gradient Boosting, Support Vector Machine, or Linear discriminant analysis (LDA) can be used. However, the present invention is not limited thereto, and an appropriate method may be selected in consideration of features of analysis data or ease of cluster prediction.

Looking at one specific embodiment, the machine learning may use a K-nearest neighbor algorithm. For example, for unknown green coffee beans, three analysis data closest to the chemical characterization data for unknown green coffee beans are selected from the chemical characterization data performed in each step, and the selected three data sets are selected. It can be predicted as a cluster that contains a majority of data among the included clusters.

According to another exemplary embodiment of the present application, a method for providing a roasting profile of green coffee beans includes the steps of: providing analysis data by analyzing chemical characteristics of green coffee beans; providing dimension-reduced data by reducing the dimension of the analysis data; forming a cluster by analyzing the reduced data; and providing a roasting profile for the formed cluster.

The steps of providing the analysis data and the dimension-reduced data and the steps of forming the cluster are the same as described above, and a description thereof will be omitted.

The term “green coffee roasting profile” may be defined as a description for producing coffee beans having a desired flavor using green coffee beans. In order to produce green coffee beans as coffee beans, three sections are passed: a drying section (input temperature ~ about 155 ℃), an intermediate section (155 ℃ ~ 200 ℃), and an expression section (200℃ ~ discharge temperature). The intermediate section may be divided into a Mayar section (155 ℃ ~ 180 ℃) and a caramelization section (180 ℃ ~ 200 ℃) again.

The drying section mainly refers to a section reaching from the input temperature to about 155° C. before the Mayar reaction occurs. In the drying section, the moisture present in the green coffee beans is converted to water vapor and is trapped inside the green coffee beans. During this section, the volume of the green coffee beans increases up to twice, the moisture content gradually decreases, and the density and overall weight of the coffee decrease. starts to become

The intermediate section means a section corresponding to about 155 °C to 200 °C, more specifically, a section corresponding to 155 °C to 180 °C means a Mayar section, and a section corresponding to 180 °C to 200 °C is a caramel section means Most of the heat applied in this section is used to increase the temperature of the green coffee beans. On the other hand, the flavor can be controlled by adjusting the length of the Mayar section and the caramel section and the amount of heat input. For example, if the Mayar section is short and the caramel section is long, acidity may decrease and sweetness may increase. In addition, when the length of Mayar is long and the caramel section is short, sweetness may decrease and bitterness may increase. In addition, if the Mayar section and the caramel section are similar, acidity and weight may be appropriately felt.

The expression section means a section from 200 °C to the discharge temperature. In this section, the thermal decomposition reaction occurs, and the Mayar reaction and the caramel reaction do not occur. In this section, the first crack occurs in which the green coffee beans are physically split without overcoming the internal pressure at about 203°C. It is common to give 15-25% of the total roasting length as the period from the start of the first crack to the discharge (Development Time Ratio, DTR), and by adjusting this length, light roasting, medium roasting, and dark roasting are determined.

In the roasting of green coffee beans, the flavor of the beans can be adjusted by adjusting the input temperature of the four sections and the time required for each section.

Various roasting profiles can be designed by setting the green bean input temperature for each classified cluster and the rate of change of temperature according to time for each section. In addition, it is possible to evaluate the flavor of coffee beans manufactured according to the variously designed roasting profiles, and obtain data on the results. Therefore, it is possible to provide a roasting profile for a flavor that can be implemented for each cluster based on the result data on the flavor.

As an example, the step of providing a roasting profile for the formed cluster may include measuring, normalizing, or normalizing the chemical component content of the formed cluster, and then, based on the measured, normalized, or normalized chemical component, the roasting profile of the green coffee beans. It may include the step of designing.

The chemical components of the green coffee beans may refer to, for example, all of the chemical components constituting the green coffee beans. As another example, the chemical component of green coffee beans may mean a chemical component having a characteristic distinguishing it from other green coffee beans, and the chemical component may be one or more. As another example, the chemical component of green coffee beans may mean a chemical component that affects the flavor of coffee.

As an example, the chemical component may include carbohydrates (monosaccharides, disaccharides or polysaccharides), organic acids, chlorogenic acids, nitrogenous compounds (proteins, amino acids, trigonelline or caffeine), lipids (triglycerides, fatty acids or diterpenes) or water. . The compounds may be changed through a series of chemical reactions (Mayar reaction, caramel reaction, and thermal decomposition reaction) as the green coffee beans are roasted, and may affect flavor.

There are various organic acids in green coffee beans before roasting, and the representative organic acids are malic acid and citric acid. Most of it is decomposed by heat during roasting, but some remains and can affect the acidity of the coffee.

Sugar is a disaccharide made by combining two monosaccharides. During thermal decomposition, it is decomposed into monosaccharides to produce various flavors.

Monosaccharides are the most basic units of carbohydrates, including glucose and fructose. As a substance that mainly participates in the Mayar reaction and caramel reaction, which are the most important reactions in coffee, it participates in all chemical reactions during the roasting process and contributes to creating various flavors of coffee. In particular, it contributes to the sweetness of my coffee by creating a sweet flavor through the caramel reaction.

Polysaccharides are complexly linked monosaccharides and exist as cellulose and mannan constituting the cell wall in green coffee beans. Polysaccharides are hardly decomposed during the roasting process and do not directly contribute to the flavor of coffee as they are insoluble in water. However, during thermal decomposition, some of the polysaccharides are decomposed to form monosaccharides, which can indirectly affect the flavor of coffee. In addition, some of the water-soluble polysaccharides remaining after decomposition may be extracted into water and contribute to the weight of coffee.

Most of the lipids in green beans exist in the form of triglycerides. It is stable against thermal decomposition during the roasting process, and the flavor created by moving from the inside of the green beans to the surface during roasting can be maintained constant. However, the flavor of coffee may change due to oxidation of lipids during storage of the beans. In particular, diterpene, a type of lipid, is not stable to thermal decomposition, unlike triglycerides, and combines with amino acids during roasting to create fruity and malt flavor.

Proteins form compounds with various flavors through Mayar reaction and Straker decomposition reaction during roasting.

Trigonelline is a kind of nitrogen compound in green beans, and it is a compound that helps antioxidant action. As the roasting progresses, it is decomposed by heat to create a sweet flavor.

Chlorogenic acid is a representative precursor of a compound that gives coffee a bitter taste. It is decomposed by heat to produce a smoky flavor compound.

Caffeine is known to be a representative substance for the bitter taste of coffee, but in reality it does not significantly affect the bitter taste.

Chemical components to be measured in one embodiment are trigonelline, glucose, fructose, other monosaccharides, sugar (sucrose), polysaccharides (Poly sugar), citric acid (Citric acid), malic acid (Malic acid) ), Other Organic Acids, Chlorogenic Acids, Proteins, Caffeine, Caffeine Sub, Lipids, Diterpenes, and/or Water not limited

As a method for measuring the content of the chemical component, a known method may be used without limitation. On the other hand, representative of chemical components, for example, if there are 7 clusters formed, the values corresponding to 7 clusters for each chemical component are ranked in ascending order, and the chemical component with the highest content is given 7 points, and the chemical component with the highest content is given 7 points. Less chemical components can represent the content of chemical components by giving 1 point. Also, known methods for normalizing data, such as min-max normalization or Z-score normalization, may be used, but the present invention is not limited thereto. In one embodiment, the normalization of the chemical component content may be normalized by converting it to a value between 0 and 1 according to the following general formula (1). In addition, using this, it is possible to obtain a normalized average value of the chemical component content for each cluster.

<General formula 1>

(Measured value of chemical component content by cluster - minimum value of population) / (maximum value of population - minimum value of population)

In Formula 1, the maximum value of the population means the highest value among the content values of the specific chemical component measured in the population, and the minimum value of the population means the lowest value among the content values of the specific chemical component measured in the population.

For example, if the population is 54 green coffee beans and a specific chemical component is glucose, the normalized value for cluster 1 is the highest among the glucose content values for 54 green coffee beans as the maximum value of the population, and the lowest content value can be taken as the minimum value of the population. On the other hand, the measured value of the chemical component content for each cluster may be the content value of the coffee glucose included in the cluster 1.

The normalized average value for the chemical content of cluster 1 is obtained by adding up all the normalized values of the chemical component content for each coffee included in cluster 1 measured in the same manner as above, and dividing by the number of coffees included in cluster 1 can

Coffee beans with excellent flavor by identifying the reaction temperature and reaction rate of chemical components for each cluster, and designing a roasting profile by setting the green bean input temperature and temperature change rate according to time according to the content of representative or normalized chemical components for each section can provide Specifically, because it provides a roasting profile for each cluster based on chemical characterization data, it is possible to predict the flavor more objectively and provide coffee beans with excellent flavor. In addition, when chemical properties of green coffee are analyzed by near-infrared spectroscopy (NIRs), it is possible to provide a roasting profile of green coffee non-destructively and quickly.

A method for predicting the flavor of coffee raw materials according to another embodiment of the present application includes providing analysis data by analyzing chemical properties of green coffee beans or coffee beans; providing dimension-reduced data by reducing the dimension of the analysis data; forming a cluster by analyzing the reduced data; measuring the chemical component content of the formed cluster; measuring the content of chemical components in the coffee raw material sample; and comparing the content of the chemical component of the coffee raw material sample with the content of the chemical component of the cluster.

The steps of providing the analysis data and the dimension-reduced data, forming the clusters, and measuring the content of the clusters and chemical components of any coffee sample are the same as described above, and a description thereof will be omitted.

As an example, the flavor prediction of coffee raw materials includes a flavor prediction of green coffee beans and a flavor prediction of coffee beans. In one embodiment, in the step of comparing the content of the chemical component of the coffee raw material sample with the content of the chemical component of the cluster, the measured chemical component content of the cluster may use the average value of the chemical components of all coffee raw materials belonging to the cluster. . In addition, the average value of all clusters measured in this way, that is, the average value of chemical components of all coffee raw materials belonging to all clusters may be used.

In another embodiment, in the step of comparing the content of the chemical component of the raw coffee sample with the content of the chemical component of the cluster, the comparison of the content of the chemical component of the coffee raw sample with the content of the chemical component of the cluster is performed for all coffees belonging to the cluster. The normalized average value of the chemical composition of the raw material may be used. In addition, the normalized average value of all clusters measured in this way, that is, the normalized average value of chemical components of all coffee raw materials belonging to all clusters may be used. It can be normalized to a value between 0 and 1 according to Formula 2 below. In addition, the normalized average value of the chemical component content of each cluster or of all coffee raw materials can be obtained by using General Formula 2.

<General formula 2>

(Chemical component content value of measurement target - minimum value of population) / (maximum value of population - minimum value of population)

In the general formula 2, the maximum value of the population means the highest value among the content values of the specific chemical component measured in the population, and the minimum value of the population means the lowest value among the content of the specific chemical component measured in the population, The chemical component content value of the measurement target means the content value of a specific chemical component included in the cluster to be normalized or the content value of all coffee raw material chemical components.

For example, if the population is 54 green coffee beans and the specific chemical component is glucose, the normalized values for all the coffee raw material chemical component content values, the highest value among the glucose content values for 54 green coffee beans is the maximum value of the population, , the lowest content value may be the minimum value of the population. On the other hand, the measured value may be the content value of the chemical component of the coffee raw material sample.

On the other hand, the normalized average value for the chemical component content of the 54 coffees can be obtained by substituting the glucose content value of the 54 coffees as the measured value in Formula 1 above. Specifically, obtain the normalized value of the chemical component content of 54 coffees measured according to Formula 1, add them all, and then divide by 54, the number of coffees, to obtain the normalized average value for the chemical component content of 54 coffees. .

When the amounts of organic acids and monosaccharides included in the coffee raw materials are equal to or less than the normalized average value, it can be predicted that the coffee raw materials have little acidity. In addition, when the amount of chlorogenic acid contained in the coffee raw material is higher than the normalized average value, it can be expected that the bitter taste is large.

In the case of green coffee beans, since the flavor may change depending on the roasting method, the flavor potential of the green coffee beans can be predicted, and when used together with the roasting profile providing method, the flavor can be more accurately predicted. In the case of coffee beans, a brewing method may be determined in consideration of the flavor predicted according to the present method, and when the brewing method is determined, a more accurate ring flavor may be predicted.

Predictable flavors include, but are not limited to, acidity, sweetness, bitterness, body, and aroma. The predictable acidity and related compounds are shown in Table 1 below.

acidity (Acidity)	Malic acid, Citric acid, Acetic acid, Formic acid, Phosphoric acid, Quinic acid
bitter (Bitterness)	Caffeine, Therobromine, Theophylline, 2,5-diketopiperazine, Caffeic acid, Ferulic acid, Chlorogenic acid, Lactone, Phenylindane
Sweetness	Sugar, Fructose, Glucose
mouth feel	Linoleic acid, Palmitic acid, Fatty acids (C16-C18)
weight (body)	Cellulose, Mannan, Arabinogalactan, Lipid (Linoleic acid, Palmitic acid), Protein
Sweet / Buttery	2,3-butanedione, 2,3-pentanedione, Damascenone, Pyridines
Honey / Caramel	Furfural, Vanillin, Furanone
Fruity	Acetaldehyde, Propanal, Phenylacetaldehyde
Malty	2-methylbutanal (2-methylbutanal), 3-methylbutanal (3-methylbutanal), hexanal (Hexanal)
Nutty / Cereal	Pyrroles, 2-acetyl pyrroline
Earthy / Roasty	2-ethyl-3,5-dimethylpyrazine (2-ethyl-3,5-dimethylpyrazine), 2-ethyl-6-methylpyrazine (pyrazine) (2-ethyl-6-methylpyrazine(pyrazine)), 2-methoxy -3-Isopropylpyrazine (2-methoxy-3-isopropylpyrazine)
Phenolic / Spicy	4-vinylguaiacol (4-vinylguaiacol), 4-ethylguaiacol (4-ethylguaiacol)
Ashy / Burn	Catechol, Guaiacol
Rotten / Sulfurous Scent (Putrid / Sulfurous)	Methanethiol, Dimethylsulfide

Example 1. Classification method of coffee raw materials

A total of 54 green coffee beans of different varieties, including 19 green coffee beans purchased from ALMACIELO, 1 green coffee purchased from Beautiful Coffee, and 34 green coffee beans purchased from GSC International as a sample of green coffee to be analyzed. was used. The purchased green coffee samples were stored for about 12 to 24 hours in an environment of a temperature of 20°C to 26°C and a relative humidity of 40% to 50%. Before taking NIR, each of the stored 54 green coffee beans was sufficiently mixed and the surface was lightly pressed before measurement so that there was no space on the surface of the NIR analysis device as much as possible.

Next, the NIR of the green coffee sample was measured using an NIR analysis device ('DS2500 F' of FOSS analytics beyond measure, Ltd). The measurement conditions were 4 repetitions and 32 subscans per one time. Therefore, when 200g of a green coffee sample is added and executed, 32 subscans are output at one time, and this is repeated 3 times for a total of 96 Subscans were performed and their average values were obtained as analysis data.

On the other hand, the length of the measured total wavelength is 400 nm to 2500 nm, and it is output in the form of a csv (comma-separated values) file at intervals of 2 nm. In order to reduce the variables depending on the color of green coffee beans, in the analysis, the wavelength range representing the color (400 nm to 700 nm) and the wavelength range corresponding to the boundary wavelength range (700 nm to 900 nm) among all wavelengths (400 nm to 2,500 nm, in 0.5 units) are Except, it was carried out using only the wavelength range value of 900nm to 2,500nm.

Principal component analysis (PCA) was performed with the analysis data obtained from 900 nm to 2,500 nm. Since there are 1,600 wavelength points to be analyzed in the 900 nm to 2,500 nm wavelength range, dimension reduction was required for data analysis containing high-dimensional information. 1 is an exemplary view showing the results of principal component analysis (Principal Component Analysis) with analysis data of 900 nm to 2,500 nm.

For the principal component analysis data, a cluster for green coffee beans was formed using a K-mean clustering method. For clustering analysis, the number of suitable clustering was found through Elbow, silhouette, or gap statistic method, and K-mean clustering was calculated using this. . In addition, it was confirmed that the cluster was well formed through hierarchical clustering analysis, and the significance was confirmed through the value of Approximately unbiased p-value (AU) of pvclust packages. Meanwhile, all analyzes related to clusters based on chemical properties of green coffee were performed in R. As a result of clustering for 54 green coffee samples, it could be classified into 7 clusters (C1 to C7). 2 is an exemplary view showing the results of forming clusters on 54 green bean samples in the method for classifying coffee raw materials according to the present application.

Example 2. Roast Profile Design Method

Then, the chemical component content of the cluster formed in Example 1 was normalized according to the above-mentioned general formula 1, and a normalized average value was obtained for the chemical component content of each cluster using this.

3 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 1 (C1). The overall characteristics of cluster 1 (C1) include a small amount of organic acid that can give acidity, and a relatively large amount of caffeine and chlorogenic acid that can give bitter taste. It also contains relatively high amounts of precursors of the Mayar reaction, such as sugars, monosaccharides and proteins. In addition, it is expected that the weight (body) is good because the amount of polysaccharides and lipids is high.

4 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 2 (C2). The overall characteristics of cluster 2 (C2) include a lot of organic acids, and sugar, protein, chlorogenic acid, and trigonelline are also appropriately included. In addition, polysaccharides, water content, and lipids that contribute to the feeling of weight are included at average levels, and chlorogenic acid and caffeine, which cause bitter taste, are contained relatively little.

5 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 3 (C3). Cluster 3 (C3) contains many chemical components such as sugar, monosaccharides, amino acids, and trigonelline, which are low molecular compounds related to rich flavor and acidity, but relatively few chemical components related to weight and bitterness are included. have.

6 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 4 (C4). Cluster 4 (C4) has a small amount of precursors (sugar, monosaccharide, and protein) of the Mayar reaction and caramel reaction, but the content of complex polysaccharides is relatively high. Sweetness produced due to monosaccharides generated by thermal decomposition can be expected. In addition, the content of polysaccharides and lipids is relatively high.

7 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 5 (C5). The overall characteristics of cluster 5 (C5) include the highest amount of all chemical components except for organic acids and complex polysaccharides.

8 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 6 (C6). As an overall characteristic of cluster 6 (C6), the sugar content is relatively low among the precursors of the Mayar reaction, and the amount of complex polysaccharides is relatively high.

9 is an exemplary diagram showing the normalized average value of the chemical component content of cluster 7 (C7). As an overall characteristic of cluster 7 (C7), the amount of organic acid and polysaccharide is relatively high, but the amount of precursor or low molecular weight compound of the Mayar reaction is relatively small. In addition, the amount of derivatives of trigonelline and caffeine is relatively high.

A roasting profile was designed by setting the green bean input temperature and the temperature change rate according to time for each section based on the content of chemical components normalized for each cluster, and the results are shown in Table 2.

	C1	C2	C3	C4	C5	C6	C7
input temperature	About 138℃ ~ 142℃	Approx. 153℃ ~ 157℃	Approx. 153℃ ~ 157℃	Approx. 118℃ ~ 122℃	Approx. 103℃ ~ 107℃	Approx. 118℃ ~ 122℃	Approx. 153℃ ~ 157℃
Mayar Section (M)	100 seconds	105 seconds	59 seconds	102 seconds	142 seconds	93 seconds	76 seconds
Caramel section (C)	107 seconds	98 seconds	132 seconds	89 seconds	131 seconds	106 seconds	144 seconds
Expression section (D)	111 seconds	84 seconds	214 seconds	105 seconds	125 seconds	169 seconds	74 seconds
exhaust temperature	Approx. 203℃ ~ 207℃	Approx. 204℃ ~ 208℃	Approx. 208℃ ~ 212℃	Approx. 207℃ ~ 211℃	Approx. 216℃ ~ 220℃	Approx. 218℃ ~ 222℃	Approx. 201℃ ~ 205℃
MCD rate	31%:34%:35%	37%:34%:29%	15%:32%:53%	34%:30%:36%	36%:33%:31%	25%:29%:46%	26%:49%:25%

Example 3. Validation of the roasting profile of cluster 5

The subjects were BRZ12 (Sehado, Brazil), PRU1 (Chanchamayo, Peru) and NEP1 (Shindupalchok, Nepal) coffee beans. As a result of clustering the green coffee beans by analyzing the chemical properties according to the present invention, all of them corresponded to cluster 5 (C5).

As described above, the normalized chemical composition of cluster 5 contains a lot of other chemical components (sugar, monosaccharide, amino acid, lipid, diterpene, trigonelline, chlorokenic acid, and caffeine) except for organic acids and complex polysaccharides. is characterized. Therefore, diterpenes and trigonelline can create a sweet aroma in the Mayar section, and when the expression section is long with a lot of protein, there is a high possibility that the bitter taste and miscellaneous taste increase. Therefore, in consideration of the characteristics of the chemical composition of cluster 5, the roasting profile was designed with a Mayar section of 142 seconds, a caramel section of 131 seconds, and an expression section of 125 seconds. Then, BRZ12 (Sehado, Brazil), PRU1 (Chanchamayo, Peru) and NEP1 (Shindupalchok, Nepal) were roasted by the method of the designed roasting profile, and then the specialty coffee taste was evaluated.

On the other hand, for comparison, it was designed with 120 seconds of Mayar section, 120 seconds of caramel section, and 120 seconds of expression section randomly selected by the roaster, BRZ12 (Sehado, Brazil), PRU1 (Chanchamayo, Peru) and NEP1 (Shindupalchok, Nepal) was compared with the evaluation of the specialty coffee taste of beans produced after roasting.

As a result of the blind test, in all cases of BRZ12 (Sehado, Brazil), PRU1 (Chanchamayo, Peru) and NEP1 (Shindupalchok, Nepal), the beans produced by the provided roasting profile for cluster 5 were produced according to the criteria randomly selected by the roaster. Compared to ground coffee beans, the unfavorable bitterness decreased, the sweetness increased, and the sweet aroma was increased, and among the criteria for evaluating the taste of specialty coffee, it was highly evaluated in the Clean cup (cleanness evaluation without harsh taste). On the other hand, the higher the score of the Clean cup, the better/sufficient the chemical reaction in the roasting process of chemical ingredients that affect flavor.

Example 4. Prediction of flavor of coffee beans

The aforementioned 54 green coffee samples were roasted by 30 g using a roaster (IKAWA pro v3 sample roaster) through the recommended roasting profile for each cluster. The roasted bean sample was stored at a temperature of 20 °C to 26 °C and a relative humidity of 40% to 50% for about 12 hours to 24 hours. Thereafter, the analysis was performed by grinding the grinder to a thickness of 0 with a grinder (HARIO V60 electric coffee grinder, EVCG-8B-K).

Next, it was measured using ATR-FTIR analysis equipment ('Nicolet IS50' and ATR diamond crystals accessory of Thermo Fisher Scientific, Ltd). The measurement conditions are per the sub-scan (subscan) 13 times, the resolution (resolution) 4cm ^-1, spacing data (data spacing) to 0.482cm ^-1 and the background collection (collection background) spectrum of the post 600 to 4,000 cm ^-1 range (spectral range) was repeated 10 times per sample. The output data was subjected to baseline correction through “ALS” of the rampy.baseline method of Python, and normalized using the average value of each sample to secure analysis data.

With the above analysis data, clusters for coffee beans could be formed using principal component analysis (PCA) and K-mean clustering methods, and flavor for each cluster could be predicted.

Example 5. Prediction of flavor of green beans

In Example 1, as measured by NIR values for 54 green coffee beans, the chemical components (Organic_acids, Sucrose, Mono_sugar, Proteins, Poly_sugar, Lipids, Diterpene, Water, Chlorogenic_acids, Caffeine, Caffeine_sub and Trigonelline) The content value was measured. The measured content value of the chemical component was normalized by converting it to a value between 0 and 1 according to the following general formula (2).

<General formula 2>

(Chemical component content value of measurement target - minimum value of population) / (maximum value of population - minimum value of population)

In the general formula 2, the maximum value of the population means the highest value among the content values of the specific chemical component measured in the population, and the minimum value of the population means the lowest value among the content of the specific chemical component measured in the population, The chemical component content value of the measurement target means the content value of a specific chemical component included in the cluster to be normalized or the content value of all coffee raw material chemical components. Meanwhile, the population was defined as the content of specific chemical components measured in 54 green coffee beans.

On the other hand, the normalized average value for the chemical component content of 54 coffees was obtained by substituting the glucose content value of 54 coffees as the measured value in Formula 1 above. Specifically, the normalized value of the chemical component content of 54 coffees measured according to Formula 1 was obtained, and after adding them all, the normalized average value of the chemical component content of 54 coffees was obtained by dividing by 54, the number of coffees.

Next, the chemical components (Organic_acids, Sucrose, Mono_sugar, Proteins, Poly_sugar, Lipids, Diterpene, Water, Chlorogenic_acids, Caffeine, Caffeine_sub and Trigonelline) content values were measured, and normalized by substituting the measured values of General Formula 1 above.

The normalized values of the coffee to be analyzed and the normalized average values of the population coffee are shown in Table 3 below.

chemical composition	Normalization of the target coffee (% increase or decrease compared to population coffee)	Normalized mean of population coffee (54 green beans)
Trigonelline	0.284 (-19.77)	0.354
Other monosaccharides (Mono_sugar)	0.364 (-4.21%)	0.38
glucose	0.263 (-27.95%)	0.365
fructose	0.493 (-8.02%)	0.536
sugar (sucrose)	0.27 (-17.93%)	0.329
Polysaccharide (Poly_sugar)	0.726 (+18.05%)	0.615
Other organic acids	0.447 (-21.85%)	0.572
Citric acid	0.454 (-21.18%)	0.576
Malic acid	0.454 (-21.18%)	0.576
Chlorogenic acids	0.67 (+21.38%)	0.552
Proteins	0.542 (+0.18%)	0.541
Caffeine	0.472 (+12.11%)	0.421
Caffeine sub	0.552 (+19.74%)	0.461
Lipids	0.606 (+41.26%)	0.429
Diterpene	0.36 (+10.77%)	0.325
Water	0.563 (+2.93%)	0.547

Compared with the coffee of the population, polysaccharides, chlorogenic acid, protein, caffeine, caffeine derivatives, lipids, diterpenes, and water components showed higher levels in the coffee to be analyzed. On the other hand, trigonelline, other monosaccharides, glucose, fructose, sugar, organic acid, citric acid, and malic acid showed low levels. Therefore, the coffee to be analyzed can predict the following flavor. * Compared to the population coffee, the content of other organic acids (0.447), malic acid (0.454), and citric acid (0.454) of the coffee to be analyzed is below the average, and the content of monosaccharides such as sugar (0.270), glucose (0.263), and fructose (0.493) is also average It is expected that relatively little acidity will be formed.

* Because the content of polysaccharides (0.726) is above average, the sweet taste of water-soluble polysaccharides is expected to be above average, but the content of sugar, glucose, and fructose, which are essential to form aroma, is below average, so sweet aroma will not be formed. As expected, it is expected to have an average sweetness overall.

* Since the content of chlorogenic acid (0.67) is above average, it is expected that the characteristic bitter taste of coffee is above average.

* The content of polysaccharides (0.726) and moisture (0.563) are above average, and the content of lipids (0.606) is high, so it is expected to form a rich body, and due to the amount of lipids, It is expected that a feeling of weight (body) will be mainly formed.

* There is a significant amount of protein (0.542), a component that forms the aroma, and the content of monosaccharides such as sugar, glucose, and fructose, which are crucial for flavor generation, also exists in a certain amount, so it is expected that the specific aroma of the coffee to be analyzed will be formed to some extent. it is expected On the other hand, since the content of diterpene (0.36) is above average, there will be aromas of fruity and malty, and since the content of trigonelline (0.284) is below average, sweet and buttery ( The aroma of buttery) is expected to be below average.

The present application also relates to a coffee information providing system. 10 is a block diagram showing the configuration of the coffee information providing system according to the present invention, Figure 11 is a block diagram showing the configuration of the terminal of the coffee information providing system according to the present invention. As shown in FIGS. 10 and 11 , the coffee information providing system according to the present invention is connected to at least one terminal 10 and the terminal 10 and the network 20 through a network 20 to provide information on coffee raw materials. It is configured to include a management server (30).

The terminal 10 is a broad concept including a wired terminal or a wireless terminal managed by various users, and includes a PC (Personal Computer), an IP television (Internet Protocol Television), a notebook (Notebook-sized personal computer), a tablet PC, PDA (Personal Digital Assistant), smart phone, IMT-2000 (International Mobile Telecommunication 2000) phone, GSM (Global System for Mobile Communication) phone, GPRS (General Packet Radio Service) phone, WCDMA (Wideband Code Division Multiple Access) phone, Including a UMTS (Universal Mobile Telecommunication Service) phone, a MBS (Mobile Broadband System) phone, etc., data of different terminals, a coffee information management server 30 and a coffee information providing system and signals, information, voice and video data A function for performing transmission and reception may be provided.

According to the user, the terminal 10 includes a producer terminal 11 used by a producer who produces green coffee beans at a production area, a roaster terminal 12 used by a roaster who produces coffee beans by roasting green coffee beans, and green coffee beans and / or the consumer terminal 13 used by the consumer who purchases coffee beans, but is not limited thereto.

The network 20 is a communication network capable of providing high-capacity, long-distance voice and data services, and may be a next-generation wired or wireless network for providing Internet or high-speed multimedia services. When the network 20 is a mobile communication network, it may be a synchronous mobile communication network or an asynchronous mobile communication network. As an example of the asynchronous mobile communication network, there may be a wideband code division multiple access (WCDMA) type communication network. In this case, although not shown in the drawing, the network 20 may include a Radio Network Controller (RNC). Meanwhile, although the WCDMA network is taken as an example, it may be a 3G LTE network, a 4G network, a 5G network, a next-generation communication network, or other IP-based IP networks. The network 20 serves to mutually transmit signals and data between each user terminal 10 , the coffee information management server 30 , and other systems.

The coffee information management server 30 is connected to a plurality of terminals 10 through a network, and when a request for information of coffee raw materials is input from the terminal 10, various information about the corresponding coffee raw materials to the terminal 100 is provided. It is provided by searching the database.

When the user requests information on raw materials for coffee through the terminal 10, the raw materials for coffee may include various types of green coffee beans or coffee bean varieties, harvest time of green coffee beans, origin of green coffee beans, producer of coffee beans, production time, or production region. You can specify the coffee ingredient by entering the information.

12 is a block diagram showing the components of the coffee information management server 30 of FIG.

First, referring to FIG. 12 , the coffee information management server 30 includes a transceiver 31 , a controller 32 , and a database 33 .

The transceiver 31 transmits and receives signals and data through a wired and/or wireless communication method with each of the user terminals 11 , 12 , and 13 through the network 20 .

The database 33 may refer to a functional and structural combination of software and hardware for storing information. The database 33 may be implemented as at least one table, and may further include a separate database management system (DBMS) for searching, storing, and managing information stored in the database 33 . In addition, the database 33 may be implemented in various ways, such as in the form of a linked-list, a tree, and a relational database, and includes all data storage media and data structures capable of storing corresponding information. .

The control unit 32 includes a coffee raw material classification information providing module 32b, a user management module 32a, a roasting profile information providing module 32c, a coffee raw material characteristic information providing module 32d, a coffee raw material flavor information providing module 32e, and a purchase management module 32f. The control unit 32 may further include a community formation and management module (not shown).

And, in this specification, a module may mean a functional and structural combination of hardware for carrying out the technical idea of the present invention and software for driving the hardware. For example, the module may mean a logical unit of a predetermined code and a hardware resource for executing the predetermined code, and does not necessarily mean physically connected code or one type of hardware. It can be easily inferred to an average expert in the technical field of

The coffee raw material classification information providing module 32b receives a request for information on coffee raw materials from the terminal connected through the network, extracts the requested coffee raw material information from the database, and controls the transceiver to transmit to the terminal . The information on the coffee raw material is classification information of the coffee raw material classified by analyzing the chemical characteristics of the coffee raw material. The chemical properties of the coffee raw material may include the type and content of a chemical element, a chemical functional group, or a chemical component (compound) contained in the coffee raw material.

The classification information of the coffee raw material may be formed by the method of classifying the coffee raw material described above, and a description thereof will be omitted.

The control unit of the present application receives a request for information on coffee raw materials from a terminal connected through a network, extracts the requested roasting profile information of the coffee raw materials, and adds a roasting profile information providing module for controlling the transceiver to be transmitted to the terminal can be included as

The roasting profile information is characterized in that it is formed based on the classification information of the coffee raw material. The roasting profile of green coffee beans may be formed by the above-described method of providing a roasting profile of green coffee beans, and a description thereof will be omitted.

The control unit of another embodiment of the present application receives a request for information on coffee raw materials from a terminal connected through a network, extracts the requested characteristic information of coffee raw materials, and controls the transceiver to transmit to the terminal characteristic information of coffee raw materials It may further include a provision module. The characteristic information of the coffee raw material includes the type and content of chemical components included in the coffee raw material.

Producers of coffee raw materials, especially producers of green coffee beans, rely only on cupping results, which are subjective judgments of a few experts, on the characteristics of their produced coffee beans or green beans. Therefore, you will want to be provided with objective and quantitative data on the characteristics of coffee raw materials, such as, for example, the content of each chemical component.

To this end, the producer of coffee raw materials may request a specific testing institution to analyze the ingredients, but in this case, only the characteristic information of a single coffee raw material can be obtained, so its utilization is limited. According to an embodiment of the present application, the characteristic information of the coffee raw material is characterized in that it is provided simultaneously with the requested coffee raw material and the different characteristic information of the coffee raw material.

In one embodiment, when the requested coffee raw material is green coffee of variety A grown in Brazil in 2020, different coffee raw materials may be selected in consideration of the variety, production region and/or production time of the requested coffee raw material.

For example, for different coffee sources, select cultivar A produced in the same year on different continents and compare the two data, select cultivar A grown in Brazil in 2015 and compare both data, or compare data from both cultivars in Brazil in 2020. You can compare the data of both by selecting the B cultivar grown in .

The control unit of another embodiment of the present application receives a request for information on coffee raw materials from a terminal connected through a network, extracts the requested flavor prediction information of coffee raw materials, and controls the transceiver to transmit the requested coffee raw material flavor to the terminal It may further include an information providing module.

The flavor prediction information of the coffee raw material may be formed by the above-described flavor prediction method, and a description thereof will be omitted.

The control unit according to another embodiment of the present application may further include a user management module for receiving and managing login information of a user from a terminal accessed through a network.

The control unit of another embodiment of the present application further includes a purchase management module for controlling the transceiver to receive payment information of coffee raw materials from a terminal connected through a network, perform a requested payment process, and transmit the result to the terminal can do.

Claims (21)

1. providing analysis data by analyzing chemical properties of green coffee beans or coffee beans;

   providing dimension-reduced data by reducing the dimension of the analysis data; and

   A method of classifying coffee raw materials comprising the step of forming a cluster by analyzing the reduced data.
2. The method according to claim 1, wherein the method of analyzing the green coffee beans or the chemical properties of the coffee beans is a spectroscopic analysis method.
3. According to claim 1, wherein the step of providing the dimensionally reduced data is a method of classifying coffee raw materials using a linear dimensionality reduction method (Linear Dimensionality Reduction Methods) or a non-linear dimensionality reduction method (Non-Linear Dimensionality Reduction Methods).
4. The method of claim 1 , wherein the providing of the dimensionally reduced data comprises Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Factor Analysis, and Singular Vector Classification of coffee ingredients using Decomposition (SVD) or t-distributed Stochastic Neighbor Embedding (t-SNE).
5. The method of claim 1 , wherein the forming of the cluster uses a supervised learning method or an unsupervised learning method.
6. According to claim 1, wherein the step of forming the cluster partitioning (partitioning), K-means (K-means), K- representative object (K-medoid), CLARA (Clustering Large Applications) or CLARANS coffee using the method How to classify raw materials.
7. The classification of coffee raw materials according to claim 1, further comprising predicting clusters through machine learning based on green coffee beans or chemical characteristic analysis data of coffee beans, and dimensionality reduction data of the analysis data. Way.
8. The method of claim 7, wherein the machine learning is a K-nearest neighbor algorithm, logistic regression, naive Bayes classification, stochastic gradient descent. , using Decision Tree, Random Forest, AdaBoost, Gradient Boosting, Support Vector Machine, or Linear discriminant analysis (LDA). How to classify coffee raw materials.
9. Forming clusters of green coffee beans according to the classification method of any one of claims 1 to 8; and

   and providing a roasting profile for the formed cluster.
10. 10. The method of claim 9, wherein providing a roasting profile for the formed cluster comprises:

    measuring the chemical component content of the formed clusters; and

    and designing a roasting profile of green coffee beans based on the measured content of the chemical component.
11. The method of claim 10 , wherein the chemical components include monosaccharides, polysaccharides, lipids, organic acids, proteins, and moisture.
12. Forming a cluster of coffee raw materials according to the classification method according to any one of claims 1 to 8;

    measuring the chemical component content of the formed cluster;

    measuring the content of chemical components in the coffee raw material sample; and

    A method for predicting the flavor of a coffee raw material comprising the step of comparing the content of the chemical component of the coffee raw material sample with the content of the chemical component of the cluster.
13. In the coffee information providing system comprising a coffee information management server and at least one terminal capable of accessing the coffee information management server through a network,

    The coffee information management server receives a request for information on coffee raw materials from a terminal and a transceiver for transmitting and receiving signals or information, a database for storing coffee information, and the terminal connected through a network, and a control unit equipped with a classification information providing module of coffee raw materials for controlling the transceiver to extract information about the data from the database and transmit it to the terminal,

    The information on the coffee raw material is a coffee information providing system that is classification information of the coffee raw material classified by analyzing the chemical characteristics of the coffee raw material.
14. The system for providing coffee information according to claim 13, wherein the chemical properties of the coffee raw material include the type and content of a chemical element, a chemical functional group, or a chemical component included in the coffee raw material.
15. 15. The method of claim 13, wherein the control unit receives a request for information on coffee raw materials from a terminal connected through a network, extracts roasting profile information of the requested coffee raw material, and controls the transceiver to transmit the roasting profile information to the terminal. A coffee information providing system further comprising a module.
16. The coffee information providing system according to claim 15, wherein the roasting profile information is formed based on classification information of coffee raw materials.
17. According to claim 13, wherein the control unit receives a request for information on coffee raw materials from a terminal connected through a network, extracts the requested characteristic information of coffee raw materials, and controls the transceiver to transmit the requested coffee raw material characteristic information to the terminal A system for providing coffee information, further comprising a providing module, wherein the characteristic information of the coffee raw material is the type and content of a chemical component included in the coffee raw material.
18. The coffee information providing system according to claim 17, wherein the characteristic information of the coffee raw material is provided simultaneously with the characteristic information of the coffee raw material different from the requested coffee raw material.
19. The flavor of coffee raw materials according to claim 13, wherein the control unit receives a request for information on coffee raw materials from a terminal connected through a network, and controls the transceiver to extract and transmit the requested flavor prediction information of the coffee raw materials to the terminal. Coffee information providing system further comprising an information providing module.
20. The coffee information providing system according to claim 13, wherein the control unit further comprises a user management module for receiving and managing login information of a user from a terminal accessed through a network.
21. The method according to claim 13, wherein the control unit further comprises a purchase management module for controlling the transceiver to receive payment information of coffee raw materials from a terminal connected through a network, perform a requested payment process, and transmit the result to the terminal coffee information providing system.

Images