EduRead

EduRead is a platform aimed at enhancing reading/writing skills among students, which employs AI technology to evaluate students' reading levels through audio recordings of reading exercises, and then provides personalized learning plans tailored to each student's proficiency level. 

Additionally, parents can register on the platform to track their children's progress and receive insights to support their development. It will include a summary of the exercises the child did and the mistakes she/he made, furthermore, EduRead will feature exercise pages designed to reinforce the specific abilities where students struggled from the target areas identified through the AI analysis.

EduRead constructs datasets using audio-text pairs, deducting points for various reading problems to categorize students. It identifies specific reading problems such as slow reading pace, lack of fluency in complex words, syllabic reading, phonological errors, and omissions or additions. 

• Low Proficiency Level: Scores ranging from 0 to 10 points.

• Medium Proficiency Level: Scores ranging from 11 to 15 points.

• Normal Proficiency Level: Scores ranging from 16 to 20 points.

Therefore, EduRead offers learning plans for each level. Low performance plans focus on building vocabulary and grammatical foundations, while medium performance plans assist with sentence structuring and fluency. Both follow the Doman Method principles and are divided into five phases. Normal performance plan focus on reading comprehension.

In the initial phase of the Low Performance Level, termed Vocabulary Building, students acquire 200 new words over 40 days. Five new words are introduced daily, with each word reviewed three times per day. Interactive exercises reinforce retention, incorporating audio recordings for pronunciation evaluation using AI speech recognition technology. Throughout this phase, students progress through a structured learning schedule, the methodology emphasizes consistency and repetition. Moving into the second phase, Word Pairs, students utilize the words they've learned to form pairs and associations. Exercises focus on relating nouns to articles and adjectives. Subsequent phases—Simple Sentences, Sentences, and Stories—introduce complex language structures and narrative comprehension activities. 

The medium performance plan adopts a progressive approach also inspired by the Doman Method, organized into five sequential phases. Initially, students engage in a 5-Day Word Sampling phase, Following this, the Word Pairing phase ensues, where students relate nouns to articles and adjectives within a week-long span. Moving forward, the Introduction to Syntax phase introduces students to basic sentence construction using nouns, articles, verbs, and adjectives, gradually building their understanding of language structure. Finally, students culminate their journey by applying their acquired skills to construct cohesive narratives in the Building Short Narratives phase.

The Normal Performance Level is guided by the VLP (Vocabulary, Lexicon, and Prediction) method to improve reading comprehension, beginning with a puzzle the student has to create and the selection of 10 to 15 key words from a given reading in order to create predictions with these materials. Upon reading, students contrast their predictions with the actual content, and then, the reading is conducted by the teacher. Further reinforcement comes through vocabulary-building exercises, such as defining words, and completing didactic exercises.

EduRead is designed to have a profound impact on children’s educational journey, mostly for those who have presente difficulties to access a high or standard education that allows them to possess their basic writing/skills. Actually, it has been proved by an expert of the field that this type of system and the methodology used can help children with reading disorders to improve their skills. By integrating methodologies such as the Doman Method and VLP method into the platform, we provide a structured and engaging approach to enhance reading and language skills.

Through activities designed to reinforce vocabulary, syntax, and comprehension, students can progress at their own pace, building a solid foundation in literacy. The utilization of AI technology for pronunciation assessment and personalized tracker ensures that students receive the period of time they need to improve their writing/reading skills. The exercises are established to last approximately a month if answered correctly; otherwise, the AI will extend the time by one or two weeks depending on the right and wrong answers of the child. In this way, students receive targeted support and additional practice where necessary, fostering a deeper understanding and mastery of language skills.

Furthermore, EduRead facilitates collaboration between educators and students, allowing educators to track students' progress and provide additional support where needed. By collecting data and insights on their children’s reading levels, we are able to provide educators or parents with progress reports detailing the advancements their children have made. (This consideration is crucial as each child's learning style is unique, and EduRead may not fully address every skill that requires improvement). Therefore, the information provided will cover individual strengths, specific skill development needs, progress tracking metrics, learning style analysis, personalized learning recommendations and educational milestones, to lastly create new exercises based on the types of mistakes the child made during the process of learning.

As EduRead is an individual project, the platform I'm developing is meaningful because it aims to share one of my greatest passions: reading. While many see books as mere papers with words, I love reading, and I can assure you that when you enjoy reading, every page becomes a portal. This portal can transport you to different and diverse worlds every single day, where I'm excited to get lost as the protagonist of a romantic comedy or where I can be the scientist who cures a strange disease. However, Peruvian teenagers do not read, and one of the factors attributed to this is the lack of skills they need to do so. The Peruvian educational system is worrying because to receive a good quality education, you have to pay for it. I have seen cases where girls and boys of 12 years old don't know how to write or read syllables.

As is known, reading develops in a person's life as a habit that influences academic success, literacy, and individual well-being. Thus, its promotion from early ages is crucial for developing comprehension and critical thinking skills. Unfortunately, last year's National Reading Survey indicated that the national literate population aged 18 to 64, on average, reads only 1.9 books per year. While this is higher than the 2019 statistics at 0.86%, it is still very low. But definitely, the most concerning situation is found in United Nations reports, where it is stated that nearly one billion children worldwide have lost basic reading and writing skills due to school attendance disruptions during the pandemic. Consequently, negative impacts on cognitive, emotional, and social development have been observed in children, affecting their ability to reach their potential and succeed in life.

In the Peruvian context, this situation is only worsening, especially for those living in rural areas, where there is a gap in education access compared to high-resource students. Besides these deficiencies, we face a decrease in the number of illiterate people in the country. According to the National Household Survey, literacy coverage in 2017 reached 94.1%; however, considering that the COVID-19 pandemic had negative consequences in the educational field, this figure has regressed. Only until 2021, the government revealed that there were about 2 million illiterate people in the country, a number primarily addressed by the program proposed by the Ministry of Education in 2006, "The Literacy Program."

Therefore, EduRead is born from this problem. My enthusiasm arises when I saw the reality of my country, and I'm aware that Latino children and teenagers face similar difficulties when they go to another country. That's why I decided to create EduRead. I have been receiving feedback from teachers specialized in the communication area, and they have told me so far that the prototype could also enhance the abilities of those children who have reading disorders.

For the moment, I've built the theme of the implementation of the methodologies to be used in the web platform, as these have to be reviewed by experts in the area. I'm currently part of the Teach For All national youth network, so I have been able to contact professionals who specialize in those areas. As I mentioned, they have provided me with the feedback that I have been incorporating to the page and on the other hand, they have told me that what I've proposed can indeed improve reading and writing skills and even help children with reading disorders.

EduRead revolutionizes traditional approaches to literacy education through its innovative use of AI technology and personalized learning strategies. The main goal is to use artificial intelligence to accurately assesses students' reading levels through audio recordings, enabling the creation of  learning plans that address individual proficiency levels and learning needs. It is not just a platform where every exercise is boring and children would not want to do the exercises, it is incorporated games to attract children and teenagers. It is also not a platform where you only will answers questions, but a website where your critical thinking will be improved. Its personalized approach ensures that students receive targeted support and intervention, maximizing their learning outcomes. Moreover, EduRead fosters collaboration between educators, students, and parents, providing insights into students' progress and offering additional support where needed. Its web-based platform enhances accessibility, enabling students from diverse backgrounds to access high-quality literacy education. With adaptive learning strategies that dynamically adjust instruction based on students' performance, EduRead ensures that each student receives the appropriate level of challenge and support, making it a pioneering solution in the field of literacy education.

Functionality of the AI system: 

Evaluates and measures a student's reading fluency by analyzing audio recordings of their reading exercises. Based on identified reading problems, the system classifies students into a determined level. Rather than displaying these levels directly, the system offers tailored recommendations to improve literacy skills and assists parents/teachers in delivering personalized education.

Dataset Construction: The dataset is constructed using audio recordings from reading exercises performed by students in public and private schools. Each audio is paired with a corresponding text. The system deducts points from the total score for identified reading problems, categorizing the student into a reading level at the end of the evaluation. (Beginner, Medium or Normal). Recommendations and designated topics corresponding to each reading level are provided in order to assist them with specialized instruction.

Input: Audio recordings of students' reading exercises

Output: Based on the identified reading issues, the system provides a recommended study plan for improvement without displaying the student's reading level.

Evaluation Criteria (X):

• Very slow reading

• Lack of fluency in complex words

• Syllabic reading

• Phonological equivalence errors (e.g., reading "seso" instead of "queso")

• Omitting or adding words

Label (Y): The reading level categorization (low, medium, or normal) based on the audio evaluation.

Level of categorization:

Based on a 20 point-score: 

Low Reading Level

Medium Reading Level

Normal Reading Level

0-10 points

11-15 points

16-20 points

The student struggles with fluency and comprehension, making multiple errors across various criteria.

The student shows moderate fluency with noticeable errors in one or more areas.

The student demonstrates strong fluency with minimal errors.

Points subtracted according to:

1. Very slow reading:

   - Deduction: 5 points

   - Explanation: If the reading pace is considerably slower than the average for the grade level.

2. Lack of fluency in complex words:

   - Deduction: 4 points

   - Explanation: Struggling with the pronunciation or fluency of more advanced or complex words.

3. Syllabic reading:

   - Deduction: 5 points

   - Explanation: Reading word by word rather than in phrases or sentences.

4. Phonological equivalence errors (e.g., reading "seso" instead of "queso"):

   - Deduction: 3 points

   - Explanation: Mispronouncing or substituting words due to phonological errors.

5. Omitting or adding words:

   - Deduction: 3 points

   - Explanation: Skipping words while reading or adding extra words not present in the text.

Note: In case of reading disorders (e.g., alexia, dyslexia, hyperlexia), an expert has mentioned that the methodologies that will be integrated in the platform could help in the learning of the children who possess reading disorders.

Therefore, EduRead uses mainly Machine Learning. Additionally, it would also employ Speech Recognition because in the exercises in the Doman and VLP Method, this technology is needed. 

If you want to have every executive summary of the project, pls write me an email to send all of them. 

I'm aware this technology works since I took a course with the U.S fellowship Equitech Futures and  my project consisted of developing an AI Blueprint. In this case EduRead was my project ( at that with a different name), therefore, I can assure that this AI system can be created. Obviously, it is audio recognition, considerations must be taken to enhance equality and no bias. 

I'm attaching the link of my project and the feedback I received.

Document

In addressing equity and combating bias in the implementation of AI within EduRead, we prioritize proactive measures to mitigate potential risks and ensure fair and inclusive outcomes. One key risk involves accent variations that may impact the AI's efficiency. To address this, we collect training data from students across different regions, encompassing a diverse range of accents. By incorporating such diverse datasets, we aim to enhance the AI's ability to accurately assess and accommodate various accents, thus promoting equitable access to educational resources for students from diverse linguistic backgrounds. Additionally, we recognize the risk of inconsistencies in voice tone leading to inaccuracies in assessment. To counter this, we train the AI to recognize and adapt to various voice tones, ensuring that assessments remain reliable and unbiased across different speaking styles. Furthermore, we acknowledge the potential interference of background noise with accurate problem identification during readings. To mitigate this risk, we implement robust noise filtering mechanisms to ensure precise problem identification, thereby promoting fairness and equity in the assessment process. Through these proactive measures, we strive to foster an inclusive learning environment within EduRead, where all students have equal opportunities to thrive and succeed, regardless of linguistic or environmental factors.

So far, I'm the only person who is working in this project. However, I'm planning to contact others software engineers to help me build this project.