No abstract

No abstract

When the Philippine government legislated that the Baybáyin should become the Philippines ' national writing system, educational institutions and other cultural organizations started planning to incorporate this writing method into the existing educational curriculum. One way back to implement this writing method is by using a smartphone app. As of October 2018, an eLearning app named Learn Baybáyin has been in the Android Playstore. This eLearning app will teach anyone how to read and write the script for Baybáyin. This mobile app is powered by a convolutional neural network (CNN) and was trained with 94% accuracy on the validation data. This study aims to replace the existing CNN model with an inception network because this type of neural network can successfully classify patterns or details regardless of its size in an image. After testing an inception model five times, the average validation accuracy was 96.2% which is higher than the current CNN model. The inception network got a higher validation accuracy than the previous LSTM (92.9%) and CNN (94%) model. Because of this, the inception network will replace the current model used by the eLearning app Learn Baybáyin to recognize handwritten characters.

Online advertising has been growing steadily over the years as more and more people connect to the internet. Advertising online can reach a large crowd which is both an advantage and disadvantage because a lot of people will see products they don't want. Targeted online advertisements are gaining popularity because it only shows advertisements to people who might be interested in the product based on their preferences and online presence. But these targeted advertisements are still not enough to narrow down the users to sell products. There might be products that might not be needed by certain people at certain times thus reducing the effectiveness of the advertisement. The text analysis tool for Instagram is developed in order to narrow down the most repeated words of media posts based on the type of photo, the gender of who made the comment, and the time the comment was posted. These repeated words are then visualized using a word cloud and word histogram so that businesses can determine which and what products to advertise. A two-week advertising experiment was done wherein during the first week, a random product was advertised on Instagram while in the second week, the text analysis tool was used to determine which product to advertise. The number of sales slightly increased in week two compared to week 1 (18 items vs 16 items sold) while the number of customer inquiries dramatically increased from 92 in the first week to 157 on the second week.

No abstract