Procedure for Creating a Virtual Multibank Agent: Key words
The system created can have a system administrator that enables modifying coefficients of belonging in a given moment, adding new words to the database or getting feedback from the system.
By means of the fuzzy logic interference motor, we can obtain an estimate (degree of certainty) of the group of contents (characteristics of the banking product) to which the question refers and after repeating this process for all the levels of the hierarchy, such that the words with a higher percentage of belonging to a level are resent to the interference motor, that is, to the diffuse logic system; the interference motor will ultimately be in charge of determining which of the question type(s) the user’s question refers, showing the corresponding responses arranged based on the probability of hit inferred and that exceed a previously determined threshold.
The interference motor implemented requires inputs and a set of rules, in order for it to give back outputs. As for the rules, we will introduce inancial calculation protocols such as the following, for example:
– Calculation of the effective APR (with commissions included)
– In the case where different interest rates are established for a same product, as occurs in the structured accounts or increasing profitability accounts, the search engine protocol will include the formula for calculating the average APR for the operation.
– In the cases where the interest rate is expressed in nominal rate or annual equivalent rate, the corresponding conversions shall be established for the APR.
– Given that in the comparison of products, we must take into account that the interest rate for the operation usually depends on the period of time, the comparison of savings products will be determined based on current values and then the amounts corresponding to the different products will be calculated at the end of their corresponding periods of time.
Given the inputs, the fuzzy sets and the rules defined, the interference motor must be in charge of calculating the output of each fuzzy system that will be related to a greater or lesser degree with the user’s inquiry. The three responses corresponding to the inquiries with the highest degree of certainty would be presented to the user, arranged based on said degree of certainty. Some examples of input and output are presented below.
Question 1: I would like information about a savings product called Fixed-Term Deposit. Response: Indicate the amount that you would like to invest and the period of time for which you would like to invest (boxes for filling in the data).
Response: Currently, the highest profitability obtained on a fixed-term deposit is offered at Bank XXX, with an effective APR (commissions included) of Y%, for a period of Z months. Additional note: The interest is not paid in cash, but by awarding a product (television), as cash remuneration is considered to be subject to taxes on any profits from investment capital. If you wish to negotiate with the bank about the substitution of the product for cash, consult the office or the website:
Question 1: I have been offered a savings product called Growing Profitability Deposit and I would like to know if the offer is interesting or if you can advise me. Response: Indicate the conditions that they offered you, capital to be invested, term and interest rates applied to each period (in the box for filling in the date, the information on the interest rate will be specifically expressed as a nominal rate or as an APR; in the same manner the period will be expressed in years, semesters, quarters, etc.)
Response: The average APR for the Growing Profitability Deposit that they offered you is Y% for a period of XX months. For this duration of time, the best offer is currently at Bank X, with an average APR of Z%. read only
This work is currently in the programming phase, and we hope to soon be able to start the real functioning tests.
The development of the work was structured into two parts. In the first part, we used the same methodology of the tabu algorithm, given that the properties of this algorithm make it ideal for optimizing navigation through websites, reducing wait times and “noise” (irrelevant information) in the search for information, and as a result, we achieve a decrease in the number of users who leave the site and we improve the overall quality of the navigation experience. In the second part, we created an intelligent agent that performs the task of sweeping the Internet according to the user’s information needs, which not only enables the user to obtain the information in the shortest amount of time possible, but to also obtain a visual map of the different bank offers for a certain product, arranged according to their similarity.
Fuzzy Logic has proven to be a powerful tool for the design of the multibank intelligent agent described in this work, because it enables creating a database structured hierarchically by levels, according to inquiries in natural language, which yields good results. However, we are working on the application of other more sophisticated—but not less promising— tools, such as neuro-fuzzy systems.
The application of these technologies represents a contribution to efficiency in the sense that it enables facilitating and considerably simplifying the tasks of capturing, storing and processing information.