KR20200130767A - Method and device for evaluating whether cryptocurrency is listed on cryptocurrency market using artificial neural network - Google Patents

Abstract

The present invention relates to a method, performed by a server, for evaluating whether blockchain cryptocurrency is listed on the cryptocurrency exchange using an artificial neural network, which comprises the steps of: collecting metadata of initial coin offering (ICO) cryptocurrency; performing deep-learning on the collected metadata of the ICO cryptocurrency using a training dataset; and outputting information on ICO success based on information on the deep-learning result and external evaluation information. Accordingly, information on whether cryptocurrency is listed on the cryptocurrency exchange can be collected not only from the corresponding ICO project team, but also by crawling online information.

Description

Method of evaluating listing on cryptocurrency exchanges using artificial neural networks{METHOD AND DEVICE FOR EVALUATING WHETHER CRYPTOCURRENCY IS LISTED ON CRYPTOCURRENCY MARKET USING ARTIFICIAL NEURAL NETWORK}

The present invention relates to a method and apparatus for evaluating the success of an ICO using an artificial neural network, and more particularly, to a method for analyzing whether a new cryptocurrency exchange is listed based on previously disclosed ICO information.

Block-chain can be applied to various technical fields, but it has developed as a technology for implementing cryptocurrency (or virtual currency) as fundraising is facilitated by ICO (Initial Coin Offering) that discloses cryptocurrency. have. For example, by using blockchain technology, Bitcoin is achieving the same value as a currency even if a third party such as a central bank, which is different from ordinary currencies, does not secure the value. Prior to the invention of the blockchain, in order to create an electronic money that would replace money, it was necessary to ensure its value or to manage the transaction records of money. However, by using the blockchain, it became possible to perform the function as a real currency without the above process.

Numerous cryptocurrencies are performing ICOs by this innovativeness of cryptocurrency, and among them, the currency that has the value as a currency and activates trading on the exchange has become only a few items.

It is known that about 80% of cryptocurrencies on the market are fake, and 12% are technically poor. Also, only about 8% are listed on the exchange, and only 2-3% of the listed cryptocurrencies are functioning and actively trading.

Most of the ICOs are close to financial fraud. After the ICO, the transaction volume of the corresponding currency is below the expected value, and its value is falling to the extent that it converges to zero.

The ICO platform is a platform that can perform investment evaluation in cryptocurrency through deep learning through crowdfunding, and can play a role of helping investors invest in cryptocurrency.

However, in the past, in most cases, experts subjectively evaluated cryptocurrency numerically and then decided whether to invest in cryptocurrency.

Therefore, investors considering investment in cryptocurrency have questioned whether the ICO will be successful with only limited ICO public data such as white papers, and the ability of the project team to attract investment or the technology applied to the cryptocurrency There is a problem that it is not possible to easily grasp information about it.

Therefore, there is an increasing need for a method that can objectively evaluate the project undertaking ICO as a whole.

However, in terms of evaluation, if the subjective opinion of the evaluator is included, the objectivity of the evaluation is lowered. Therefore, for fair evaluation, it is necessary to output numerical evaluation results focusing on objective items.

The present invention is to solve the problems of the prior art described above. In the present invention, information on whether to be listed on a cryptocurrency exchange can be collected not only from the corresponding ICO project team, but also online information by crawling.

In addition, the present invention may provide objectified ICO success prediction information by deep learning the collected ICO information and external evaluation information such as experts.

An embodiment of the present invention is performed by a server, performed by a server, in a method for evaluating the listing status of a blockchain cryptocurrency exchange using an artificial neural network, comprising: collecting metadata of ICO cryptocurrency; Deep learning the collected metadata of the ICO cryptocurrency using the training dataset; And outputting information on ICO success based on the deep learning result information and external evaluation information.

Another embodiment of the present invention is a block chain cryptocurrency exchange listing evaluation server, ICO data receiving unit for collecting metadata of the ICO cryptocurrency; A learning unit for deep learning the collected metadata of the ICO cryptocurrency using the training dataset; And an ICO success information output unit that outputs information on ICO success based on the deep learning result and external evaluation information.

The present invention relates to a method of predicting whether to be listed on a cryptocurrency exchange, and before the ICO, the probability of listing on an exchange of a new cryptocurrency can be output as an objective number.

1 is a flowchart illustrating a method of evaluating whether to list a cryptocurrency exchange according to an embodiment of the present invention.
2 is a block diagram conceptually showing the structure of a server for evaluating whether a cryptocurrency exchange is listed according to an embodiment of the present invention.
3 is a conceptual diagram illustrating machine learning according to an embodiment of the present invention.
4 is a diagram for explaining a process of collecting expert evaluation according to an embodiment of the present invention.
5 is a diagram illustrating an example of information on predicting success or failure of an ICO according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement the present invention. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and similar reference numerals are assigned to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only "directly connected" but also "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

In the present specification, the term "unit" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Further, one unit may be realized using two or more hardware, or two or more units may be realized using one hardware. Meanwhile,'~ unit' is not meant to be limited to software or hardware, and'~ unit' may be configured to be in an addressable storage medium or configured to reproduce one or more processors. Thus, as an example,'~ unit' refers to components such as software components, object-oriented software components, class components and task components, processes, functions, properties, and procedures. , Subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays and variables. The components and functions provided in the'~ units' may be combined into a smaller number of elements and'~ units', or may be further divided into additional elements and'~ units'. In addition, components and'~ units' may be implemented to play one or more CPUs in a device or a security multimedia card.

The "user terminal" mentioned below may be implemented as a computer or portable terminal that can access a server or other terminal through a network. Here, the computer includes, for example, a notebook equipped with a web browser, a desktop, a laptop, and the like, and the portable terminal is, for example, a wireless communication device that ensures portability and mobility. , IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet), LTE (Long Term Evolution) communication-based terminal, smart It may include all kinds of handheld-based wireless communication devices such as phones and tablet PCs. In addition, the "network" refers to a wired network such as a local area network (LAN), a wide area network (WAN), or a value added network (VAN), or a mobile radio communication network or satellite It can be implemented in any type of wireless network such as a communication network.

1 is a flow chart showing a method of evaluating the success of an ICO according to an embodiment of the present invention. Here, the success or failure of an ICO means whether a cryptocurrency will be listed on an exchange or the value of a cryptocurrency after being listed on an exchange. It can mean both going up or going down.

As shown in Fig. 1, in order to evaluate the success of the ICO, the server collects metadata of the cryptocurrency (S110).

The server can collect metadata from various paths. It is possible to collect a large amount of data such as information of each cryptocurrency (approximately 1600 cryptocurrencies as of September 2018), white papers issued, SNS information of project team members and team members, social relations information, coin circulation period, coin market cap, etc. have.

The way the server collects metadata can vary. The server can collect it automatically through Internet crawling, and in some cases, it is possible to collect it by directly attaching a document or URL.

The server may perform deep learning on the collected cryptocurrency metadata by using a training data set (S120).

For example, the collected data is used as an input, and the output is used as a team, technical information, product, and the quantitative evaluation values of experts for feasibility. After setting, deep learning can be performed by inserting a training dataset.

In addition, as another example, it is possible to obtain an output value by adding quantitative evaluation information of an expert on cryptocurrency as an input value.

In this case, the deep learning method is a method of constructing a neural network of a computer, constructing a data set, and learning about it. That is, it may be a method of constructing a learning algorithm using all or part of the collected metadata information on the ICO cryptocurrency as a training data set and applying it to a new ICO cryptocurrency.

The server may output information on the success of the ICO based on information obtained as a result of the deep learning and external evaluation information (S130).

The server can evaluate the new cryptocurrency through deep learning and output a numerical result on whether the ICO can be successful based on external evaluation information collected from cryptocurrency experts.

2 is a block diagram conceptually showing the structure of a server for evaluating whether a cryptocurrency exchange is listed according to an embodiment of the present invention.

As shown in Figure 2, the cryptocurrency exchange listing evaluation server (200 (hereinafter,'server')) includes an ICO data collection unit 210, a learning unit 220, and an ICO success information output unit 230. I can.

The ICO data collection unit 210 may collect metadata from various paths. It is possible to collect a large amount of data such as information of each cryptocurrency (approximately 1600 cryptocurrencies as of September 2018), white papers, SNS information of project team members, team members, social relations information, coin circulation period, coin market cap, etc. .

The ICO data collection unit 210 provides basic information of cryptocurrency through a site that discloses ICO information (for example, the name of the cryptocurrency, team member information, the ICO start date and end date for the cryptocurrency, the platform on which the ICO is executed) , ICO roadmap, token type, ICO status, restricted country list, etc.), SNS information for the corresponding cryptocurrency, and white paper information for the corresponding cryptocurrency can be collected. In addition, the ICO data collection unit 210 may collect transaction information such as transaction volume, transaction amount, and transaction trend of cryptocurrency listed through the cryptocurrency transaction volume site. In addition, the ICO data collection unit 210 may collect structured data (the number of followers, the number of comments, etc.) and unstructured data (posts, images, etc.) from the SNS account for each cryptocurrency. This data collection can be performed periodically.

The ICO data collection unit 210 may collect metadata in various ways. It can be automatically collected through Internet crawling, and in some cases, it is possible to collect it by directly attaching a document or URL.

The ICO data collection unit 210 matches basic information of the cryptocurrency and the structured SNS data of the cryptocurrency and inserts it into a table in the database after performing an ICO labeling operation. In addition, the ICO data collection unit 210 collects the SNS unstructured data, inserts it into a table in the database, performs matching with the information of the ICO-labeled table, downloads the ICO white paper of the corresponding cryptocurrency, and extracts only the text. Into a separate table.

The learning unit 220 may pre-process the collected metadata. The learning unit 220 designates information on the cryptocurrency name, country, and structured data (for example, the number of likes, the number of followers, the number of comments, etc.) of the collected metadata as an immutable variable, and each password Information on the currency's ICO period, cryptocurrency project keywords, PRE ICO price, ICO price, team member list, soft cap, hard cap, minimum investment amount, tradable coins, and number of tokens to be sold can be specified as variable variables.

The learning unit 220 defines each variable as a numerical value and expresses the values of the variables. Thereafter, each variable can be defined as a value of 0 to 1 through normalization.

The learning unit 220 may learn metadata of the collected ICO cryptocurrency.

The learning unit 220 may perform learning using a statistical model. For example, learning can be performed using logistic regression techniques.

The learning unit 220 may perform deep learning on the metadata of cryptocurrency using a training dataset.

The learning unit 220 may configure an artificial neural network by setting a training data set, and may perform deep learning using the artificial neural network.

The learning unit 220 may perform deep learning in two versions.

The learning unit 220 may perform deep learning using all of the above-described variables. In this case, the weight of the meaningless variable becomes 0 as learning progresses, so that it does not affect the model.

The learning unit 220 may use four variable selection methods to prioritize variables and then average them to select several variables in the highest order, and perform deep learning based only on the selected variables.

In order to describe in more detail a method of performing deep learning by configuring an artificial neural network in the learning unit, reference will be made to FIG. 3.

As shown in FIG. 3, various nodes exist between an input layer and an output layer, and a hidden layer composed of these nodes may exist. The above-described variables may be applied to each input value and output value.

As shown in FIG. 3, an algorithm including a plurality of input nodes I1 to I5, intermediate nodes H1 to H3, and output nodes O1 will be described as an example. At this time, each node may perform an operation process of assigning a weight to an input value. Data sets including information on existing ICO cryptocurrencies may be input to the input nodes I1 to I5. In other words, it can have various input values, such as basic information of cryptocurrency (I1), white paper (I2), project team member information (I3), SNS information of team members (I4), and market capitalization of coins (I5). The example of the input node is given for convenience of explanation, and it is sufficient if it relates to information on the cryptocurrency that has performed an existing ICO.

The output value may be a numerical value for the success of the ICO. The numerical value for the success of the ICO can be calculated by the average ratings of various items. Alternatively, it can be calculated as a probability value for the success of the ICO.

Values input to the input nodes I1 to I5 may be -1 to 1. For example, if there is an ICO history among I3 (ICO project team members), it can be entered as 1, and if there is no ICO history, it can be entered as -1. In addition, of I4 (SNS information of team members), +1 if SNS is the most active (if there is an activity history above a certain level), 0 if it is a moderate level of SNS activity, -1 if the team member does not perform any SNS activity at all. Can be entered as

On the other hand, when values input to the input nodes I1 to I5 are set in the above manner, machine learning is performed so that the correct correlation can be determined by using a set of a plurality of input values-output values match. I can.

For example, assuming that input values corresponding to I1 to I5 are one set, and defining five sets including different input values, the first set, the second set, the third set, the fourth set, and Assume that the output values for the fifth set are 0.1, 0.2, 0.3, 0.4 and 0.5, respectively. At this time, the sum of the output values can be 1.5.

The learning unit 220 may perform correction on the output value of 1.5. Specifically, 1.5 may be a value collected through a specific team member. However, since the ICO project team may have several team members, correction work is required to generalize this. Therefore, the value obtained by dividing the sum of the SNS activity index for all team members by the number of team members can be regarded as the SNS activity value of the team member.

Meanwhile, a weight of -1 to 1 may be additionally assigned to each node to be calculated. In addition, at least one of four arithmetic operations (addition, subtraction, multiplication, and division) of values reflecting the weights may be performed in the intermediate nodes H1 to H3.

In this case, the most important point for increasing the accuracy of the learning algorithm may be the number of intermediate nodes H1 to H3 and the size and sign (+ or -) of a weight value between nodes. In the present description, the intermediate nodes H1 to H3 are expressed as three, but may be set to a different number in some cases.

On the other hand, determining the weight between nodes can play a very important part in the accuracy of the algorithm as to which value to start with. If you start with an incorrect value, you may fall into a local minimum and you may not be able to perform weight optimization. That is, if the initial value is incorrectly set, an error in which the minimum value stays only in the local minimum value may be derived even if any initial value change is performed thereafter.

Therefore, in order to solve this problem, the learning unit 220 may utilize a genetic algorithm (GA).

For example, the learning unit 220 may repeatedly perform the aforementioned weight optimization process by generating four different sets of weights between nodes and setting each set as an initial weight value. Subsequently, the output values are compared through each set, and a set representing the smallest difference from the actual output value may be set as an initial weight value set.

The ICO success information output unit 230 may output information on whether the ICO of the new cryptocurrency will be successful. At this time, as a criterion for success, it may be listed on an exchange or set to be traded in excess of the standard trading volume.

In addition, the ICO success information output unit 230 may output evaluation result information for each item in addition to information on success or failure. For example, items such as team member items, technical information, marketability, and feasibility can be output in the form of a rating for each item. A detailed description of this will be described later.

4 is a diagram for explaining a process of collecting expert evaluation according to an embodiment of the present invention.

As shown in FIG. 4, the server may collect evaluation information on ICO from experts on ICO. Although ICO is evaluated as an objective data, information on whether there is practical investment value can be skillfully analyzed by an ICO expert, so subjective information from experts can also be an important indicator in predicting the success of an ICO.

The server may receive rating information for various items from an expert through the terminal. You can receive rating information in four categories: team, business feasibility, technology, and product. At this time, the team can be a criterion for rating detailed items on managers, developers, partners, advisors, and marketing, and business can be a criterion for rating service, marketability, advertisement, etc., and technology is performance , Verification, MVP, ethics, etc. can be the criteria for rating, and for a product, basic matters, progress, etc. can be the criteria for rating. In addition, the composition of the ICO project team members and the competency information of each team member, the team evaluation on the skills, career, and position, etc. ) Rating information can be input in items such as evaluation, investment value, scalability, marketability, and versatility of product evaluation and utilization, accessibility, accessibility, feasibility evaluation for design models, etc.

This expert's external evaluation information can be carried out in all or part of the cryptocurrency that has performed the existing ICO. As a server collecting external evaluation information, if it has a lot of external evaluation information, it can be advantageously applied to the establishment of a new ICO success prediction model, but if there is a certain amount of external evaluation information, it is considered sufficient to establish a new ICO success prediction model. I can.

This expert's external evaluation information may be applied to an input layer in the deep learning process described above. In particular, when establishing a model for predicting the success or failure of a new ICO, if external evaluation information from experts is used as an input value, a hybrid model (Hybrid Due Diligent (HDD)) combined with deep learning and expert evaluation can be implemented. Have.

The output value of the prediction of the success of such a new ICO is deep-learned along with external evaluation information from experts after a certain period of time after the ICO is made, so that information on the existing cryptocurrency can be changed, and the deep learning model can be continuously updated.

5 is a diagram illustrating an example of information on predicting success or failure of an ICO according to an embodiment of the present invention.

As shown in FIG. 5, a success predicted numerical value for a cryptocurrency to be newly ICO may be displayed through a terminal and provided to a user. The criterion of success may be the probability of trading more than the standard trading volume for a certain period, and may be expressed in the form of probability.

In addition, rating information for each item (team, technology information, productivity, and feasibility) for a new ICO may be output together. As an investor, it has an advantageous effect of being able to easily decide whether to invest in an ICO by recognizing the rating for each item or the overall evaluation information.

In addition, as a further embodiment, the ICO success information output unit 230 of the server may evaluate how much value each cryptocurrency will have after listing through ICO and provide it to the user. As described above, when the server inputs basic information of cryptocurrency, SNS data, and white paper information based on the existing learning results, the server can evaluate whether the cryptocurrency will succeed in ICO. Further, the server can learn and predict which cryptocurrency draws what value fluctuation curve after listing by learning the trend of changes in the value of the cryptocurrency that succeeded in the existing ICO over the months after listing. For example, if the listing price at the time of listing on a cryptocurrency exchange is 1.0, it is possible to predict whether the value of the corresponding cryptocurrency will be 1.5, 0.2, or 1.0 after a certain period (for example, 3 months). Through this, the server receives information about the cryptocurrency that has not yet been listed (basic information, white paper, SNS account), and then predicts how many times the value of the cryptocurrency will increase after listing, whether the value will draw a downward curve, etc. I can.

In addition, as an additional embodiment, the server may recommend cryptocurrencies for each investor. First, the server is the basic information that each investor subscribes to (age, gender, keywords of interest, etc.), the use pattern of the site evaluation site for listing on a cryptocurrency exchange (for example, the cryptocurrency that was clicked before, the order of content display), and investment in virtual currency You can determine what kind of cryptocurrency you will be interested in based on your history. After that, the server can recommend the list of ICO cryptocurrencies to investors in the order of the highest probability of ICO success (listing possibility) among the cryptocurrencies that best match the interests of the investor.

In addition, as an additional embodiment, the server may refer to texts contained in the white paper when performing deep learning to determine whether to be listed on a cryptocurrency exchange. In other words, texts frequently appearing in white papers of successful cryptocurrencies (e.g., transactions, decentralization, ecosystem, order, etc.) and texts frequently appearing in white papers of failed cryptocurrencies (e.g., users, investments, create, etc.) Price, etc.), and based on this, it is possible to consider the evaluation of the listing of cryptocurrency exchanges by determining how similar the texts frequently appearing in the white papers of the cryptocurrency to be evaluated have with the existing texts collected.

An embodiment of the present invention may also be implemented in the form of a recording medium including instructions executable by a computer such as a program module executed by a computer. Computer-readable media can be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Further, the computer-readable medium may include both computer storage media and communication media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Communication media typically includes computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave, or other transmission mechanism, and includes any information delivery medium.

Although the methods and systems of the present invention have been described in connection with specific embodiments, some or all of their components or operations may be implemented using a computer system having a general-purpose hardware architecture.

The above description of the present invention is for illustrative purposes only, and those of ordinary skill in the art to which the present invention pertains will be able to understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

200: Cryptocurrency exchange listing evaluation server

Claims (12)

In the method of evaluating the listing status of a cryptocurrency exchange using an artificial neural network performed by a server,
Collecting metadata of cryptocurrency;
Deep learning using the collected cryptocurrency metadata and ICO success results as a training dataset; And
Outputting information on success or failure of an ICO for a new cryptocurrency based on the deep learning result information;
Including,
The metadata includes basic information about the existing cryptocurrency, ICO information, and SNS information, a method for evaluating the listing of a cryptocurrency exchange using an artificial neural network. The method of claim 1,
The metadata is,
A method for evaluating the listing of a cryptocurrency exchange using an artificial neural network, characterized in that it further comprises at least one or more of a white paper of cryptocurrency, ICO team member information, and coin distribution information. The method of claim 1,
The deep learning step,
A method for evaluating the listing of cryptocurrency exchanges using an artificial neural network, characterized in that deep learning by setting different weights for each training set. The method of claim 1,
The step of outputting information on whether the ICO was successful or not,
Deep learning is performed by additionally considering external evaluation information, and information on whether to be listed on the cryptocurrency exchange is output, but the external evaluation information is,
A method for evaluating the listing of a cryptocurrency exchange using an artificial neural network, characterized in that it includes information on the ICO project team, technical information, and quantitative evaluation information of experts on marketability and feasibility. The method of claim 1,
The step of outputting information on the success of the ICO
A method of evaluating whether a new cryptocurrency is listed on a cryptocurrency exchange using an artificial neural network, characterized in that in the step of outputting information on a probability that a new cryptocurrency will be traded more than a reference transaction volume after the ICO.
The method of claim 1,
The step of outputting information on the success of the ICO
A method for evaluating the listing of a cryptocurrency exchange using an artificial neural network, comprising the step of outputting information on whether the value of the cryptocurrency in the exchange will increase after the new cryptocurrency has succeeded in the ICO.
The method of claim 1,
Based on the website usage history and cryptocurrency investment pattern of investors who have connected to the server, recommending cryptocurrencies that fit the preferences of investors, but in order of cryptocurrencies with the highest ICO success, using artificial neural networks. How to evaluate the listing of cryptocurrency exchanges.
In a cryptocurrency exchange listing evaluation server,
ICO data receiver for collecting metadata of cryptocurrency;
A learning unit for deep learning using the collected cryptocurrency metadata and ICO success results as a training dataset; And
An ICO success information output unit for outputting information on whether the ICO for a new cryptocurrency is successful based on the deep learning result;
Including,
The metadata is characterized in that it includes basic information on the existing cryptocurrency, ICO information, and SNS information, cryptocurrency exchange listing evaluation server. The method of claim 6,
The metadata is,
A cryptocurrency exchange listing evaluation server, characterized in that it further comprises at least one or more of a white paper of cryptocurrency, ICO team member information, and coin distribution information. The method of claim 6,
The learning unit,
A cryptocurrency exchange listing evaluation server, characterized in that deep learning by setting different weights for each training set. The method of claim 6,
The ICO success information output unit performs deep learning by additionally considering the external evaluation information and outputs information on whether to be listed on the cryptocurrency exchange, wherein the external evaluation information,
A cryptocurrency exchange listing evaluation server, characterized in that it contains the ICO project team information, technical information, and quantitative evaluation information of experts on marketability and feasibility. The method of claim 6,
The ICO success information output unit,
In the step of outputting information on a probability that a new cryptocurrency will be traded more than a reference transaction volume after the ICO, a cryptocurrency exchange listing evaluation server.

Images