Text Recommendation By Ogawa - Stuck for words? We've got your back

TRON Hackathon Hackathon Season 7
1

Project Name: Text Recommendation

Project Track: Web3

Team Name: Ogawa

Team Member(s): IanOgawa (Profile - ianogawa - TRON DAO Forum)

DevPost Project Link: Text Recommendation | Devpost

Project Goal: The primary goal of the Text Recommendation app is to address the common challenge of finding the right words in various communication scenarios, whether it’s for crafting a client email, initiating a chat with a colleague, or maintaining an appropriate tone. By providing AI-driven, style-specific suggestions, the app aims to reduce the anxiety and uncertainty around written communication, enabling users to convey their messages confidently and effectively. Additionally, it allows users to learn from the recommendations, enhancing their communication skills.

Project Value: Integrating the Text Recommendation system with blockchain technology ensures data integrity, security, and transparency. By leveraging blockchain, user interactions and recommendations are securely stored, preventing unauthorized changes and providing a verifiable history. This integration not only enhances trust but also adds a layer of accountability, making the app suitable for professional and business use cases where reliable communication is crucial.

Project Info:
project-info.pdf (355.3 KB)

Project Website: https://text-recommendation.onrender.com

Project Test Instructions:
Please note that due to certain limitations, the initial page load may take up to 50 seconds, and the system can handle a maximum of 15 requests per minute.

credentials.pdf (17.0 KB)
https://youtu.be/dUHPFdYfgM0

Project Details:
Overview
Text Recommendation is an app designed to help users craft professional and effective communication by providing tailored text suggestions based on sentiment analysis. It offers options in different styles (business formal, casual) and includes tones translation capabilities, enabling users to learn from the recommendations to improve their communication skills.

Features

  • Sentiment Analysis: Analyzes incoming text to understand tone and context and generates reply suggestions in different styles (formal, casual, etc.) to suit various communication scenarios.
  • Initiation Suggestions: Offers suggestions for initiating a chat or email with a chosen recipient (Client, Superior, etc.), based on the selected tone.
  • Tone Translation: Translates the tone of the text, helping users adjust their message to be more formal or casual as needed.
  • Learning Component: Enables users to improve communication skills by learning from the recommended replies.

Technology Stack
The project is built using Ruby on Rails, TronWeb, Solidity, Bootstrap, StimulusJs, Gemini-AI, and hosted on render.com.

Functionality
The app evaluates incoming texts and generates suitable responses based on sentiment and style preferences. Users receive multiple suggestions for different tones, or they can learn to craft better responses. The app also provides initiation suggestions, helping users start a chat or email with a chosen recipient in the desired tone. Additionally, it includes a tone translation feature, allowing users to adjust the formality of their text as needed. Integration with blockchain ensures data security and verification.

Integration
Text Recommendation integrates with blockchain technology using TronWeb, enhancing data security and ensuring the integrity of user interactions and communication history.

Target Audience
The app is ideal for professionals, businesses, and individuals who need assistance in crafting clear, effective, and appropriately toned communication.

Current Status
We have developed a working prototype, redesigned the app interface, and successfully integrated blockchain technology, making it ready for further testing and optimization.

What’s Next for Text Recommendation

  • Expanding query parameters to provide more versatile and tailored text recommendations, accommodating various communication scenarios.
  • Optimizing the AI model to enhance consistency and reliability in generating responses, ensuring more accurate and effective suggestions.

Smart Contract links: TRONSCAN | TRON BlockChain Explorer | 波场区块链浏览器

Project Milestones:
Milestone 1: Redesign - Update the design to improve user experience.

Milestone 2: Skill Development - Learn and enhance skills in frontend development and blockchain technology to ensure a more robust project update.

Milestone 3: Blockchain Integration - Implement TronWeb into the app to enable blockchain functionality.

26 Likes
2

Welcome to the hackathon session 7.this will be interesting project. But sir why is your project on Builder track. Is like you a new person.

Sir will your app be available for iPhone and android?

Screenshot_2024-10-06-15-20-27-655_com.brave.browser
Screenshot_2024-10-06-15-20-27-655_com.brave.browser720×1640 103 KB
sir there is no sign up, only login or I should login my normal email and password?

3 Likes
3

Hi @Relate101 Thank you.

For the track, I was initially unsure whether to submit it under Web3 or Builder, as it’s an existing project with significant updates through blockchain integration.

Yes, I plan to release it on iPhone and Android later.

As for the sign in, I’ve attached the credentials in the Project Test Instructions section. file credentials.pdf project-info.pdf

3 Likes
4

Since you are new to Tron HackaTron, it’s best to change your track to Web3. Builders track is for project that has participated in previous seasons of the hackathon and are continuing updating their projects significantly

4 Likes
5

Hi @Youngyuppie Thank you, I’ve switched it to Web3.

1 Like
6

Welcome to Grand hackathon S7
This text recommendation is it similar like cocktail messages with saved templates or does your own generate fresh words each time?

1 Like
7

Why do i need this when I can use ChatGPT to help my words, and simplify the process? and why blockchain improve my privacy? especially when blockchain itself focuses on transparency?

Instead, you actually store the text and history into blockchain state? Everyone can read each other now, so where is the privacy? You know even if it’s private variable in the contract but the raw transaction from input make it readable for everything!

4 Likes
8

Hi @Nweke-nature1.com , Thank you! My text recommendation system isn’t based on saved templates. It uses generative AI to create fresh, unique responses dynamically for each interaction. While the system might occasionally generate the same response, it is not guaranteed and varies depending on the input.

4 Likes
9

hi @theras Great question! While you can use ChatGPT for general text generation, my app is tailored specifically for providing more relevant and context-aware recommendations. This specialized focus helps users feel more comfortable, as they get tailored results suited to their needs, much like how Leonardo AI is better suited for generating images.

Regarding privacy and blockchain: Users have the option to not save the prompts or results, and there’s a clear notice to avoid sharing sensitive information in the prompts. Blockchain ensures transparency, but in this case, it’s used selectively to provide verifiable results without compromising privacy. It enhances security because users have control over what gets stored and shared.

Think of it like how ‘Gravity Write’ is specifically built for generating scripts—each tool has its unique focus and strengths. My app is designed to optimize text recommendations in a way that general tools, like ChatGPT, might not do as efficiently.

3 Likes
10

As i said, not just general text generation. You can create a specific tailored and context-aware recommendations. ChatGPT offers extensive customization through fine-tuning with user data. You can explore the tailored one from here Explore GPTs

You claim that your results are better because your app is tailored, but have you actually compared them to what ChatGPT produces? Can you give the result of the data? The training on supercomputers means ChatGPT has access to a vast amount of knowledge and context—how does your app stack up against that?

What do you mean? I want to SAVE the result, but i dont want it to be leaked randomly!

Simply allowing users to decide what to store doesn’t guarantee that their data is secure. Control over storage does not inherently ensure that the stored data is secure from breaches or misuse.

1 Like
11

Hi @theras I appreciate your insights and feedback! You’re right that ChatGPT offers extensive customization through fine-tuning, but my project uses Gemini-AI, which also leverages supercomputers to provide highly specialized recommendations. When you mention supercomputers, do you mean to suggest that Google (with Gemini-AI) doesn’t use supercomputers for its AI models?

To your point about comparing results: Yes, I have considered ChatGPT, but my app is designed to serve a very specific purpose—providing tailored recommendations based on certain user needs. While ChatGPT can be customized, having a purpose-built system often enhances user comfort, much like how tools like Leonardo AI are better for generating images. I believe it’s more about the right tool for the right job.

Because this is an existing project that uses Gemini-AI, I integrated blockchain for the hackathon to explore the added benefits of privacy, transparency, and verifiability.

As someone experienced in AI and blockchain, I would love to learn more from you. Could you explain if your stance means that all blockchain-based solutions aren’t secure? While transparency is a key feature of blockchain, many platforms (including mine) offer options to avoid saving sensitive data, and even ChatGPT itself sometimes warns users not to share such data since it’s used for training models.

I’m genuinely interested in hearing your thoughts and learning from your expertise!

2 Likes
12

I meant it same with Gemini-AI or ChatGPT both can tailored the results. Why not I just use customized Gemini-AI directly instead of using this?

So this raises the question: why do I need to add the complexity of blockchain? It doesn’t seem to solve a real problem when it could be directly customized within the Gemini AI platform. Aren’t you just extending Gemini AI and making users pay extra without providing a proper market fit?

First, you mistook the blockchain security context here. There’s a clear distinction between identity security and transaction security.

Identity Security vs. Transaction Security: Identity security focuses on protecting personal data and preferences, while transaction security ensures data integrity and immutability on the blockchain.

My concern is about identity security, while you keep referring to the security of blockchain in terms of mutability.

  • Blockchain is secure when handling data mutability, requiring more than 50% of nodes to be compromised for changes to occur.
  • Blockchain isn’t secure when it comes to data privacy.

Blockchain is primarily focused on transparency, which can create tension with privacy needs, especially when ensuring the security of communication history on the blockchain. Communication history is moderately sensitive data.

So, there is a distinction between anonymity and pseudonymity too. Many AI platforms advise against storing highly sensitive data, such as private keys and passwords, because even moderately sensitive data is often secured on centralized servers. However, your technology doesn’t securely store moderately sensitive data; instead, it exposes it to the blockchain ecosystem and all the communication history!

I give you example, I want to train the model on my project ideas and preferences. While this data isn’t highly sensitive because it’s not private-key or password and wouldn’t be a problem if it truly leaked, BUT I still want to keep it safe.

Dev can create UI that will decrypt the bytes from raw transaction inputs,
This is why subgraph exist! to simplify developer accessing the data from blockchain.
Whatever data in the blockchain, it is actually transparent.

4 Likes
13

Wow i must say this is a high level of education and enlightenment
Not only newbies is this helpful to but it’s very helpful to professionals including crypto OGs :muscle:
I have always wondered how secured is my identity on web, I just have to trust and upload my details without even knowing exactly what happens to my identity on the end which I cannot see. Over the years, blockchain technology has proven to be the benchrock we will relly on in the coming transformative world of advancing technology
Imagine a world no one holds you ransomed for anything,
It would be a free world and safe place to be
Question is, how long before we get there ?.

1 Like
14

Welcome to the Hackathon of season 7, I have read everything in you’re project, please tell me what is the long term vision for the text recommendation app, thank you

1 Like
15

Hi @theras Thank you for the great insights! As a newbie who’s just started learning about blockchain in less than a month, I’ve learned a lot from your expertise. That’s what the community is for, right?

As you’re already aware, my submission is a prototype, essentially an MVP, where the focus for this hackathon was to integrate blockchain, which is allowed by the rules. I understand this might not address all privacy concerns, but it’s an exploration of integration possibilities.

On your point about using Gemini-AI or ChatGPT directly: Does this mean that every platform integrating with ChatGPT or Gemini-AI is useless, and everyone should just use those tools directly? If so, would that imply that all hackathons using these AI models are redundant since we could just use ChatGPT or Gemini AI alone?

Thanks to your feedback, I have some ideas for future iterations, maybe for the next hackathon season. I’d love to hear your suggestions on how to address the privacy concerns you raised. It would be incredibly helpful for my learning, and I imagine you might be solving similar issues in your own apps.

After all, the community is also a space to share knowledge and grow together, right?

1 Like
16

hi @Bullish Thank you for your thoughtful insights! To be honest, as a newbie, there’s already a lot of information going through my head, and I’m still learning. The learning curve is steep, but this discussion has been really helpful.

While blockchain holds a lot of promise for securing identities and data, as you mentioned, the technology is still evolving. It’s exciting to think about the potential of a world where no one can hold your data hostage, but it seems there’s still work to be done in balancing privacy and transparency.

I really appreciate being part of these conversations—it’s a great way for me to keep learning and grow in this space.

2 Likes
17

Hi @ines_valerie thank you.

Long-Term Vision for the Text Recommendation App

  1. Multi-Platform Integration

    • Mobile App Expansion: Since I plan to implement a mobile app, the long-term goal is to create a seamless ecosystem across mobile, web, desktop, and even wearables. Users will receive real-time, personalized text recommendations on any device they are using.
    • Third-Party Integrations: The app could eventually integrate with major platforms like email clients, social media, and productivity tools, allowing users to get contextual recommendations anywhere they write.

  2. AI-Powered Contextual Understanding

    • Deeper Contextualization: The AI model will evolve to better understand the context, tone, and style preferences of users. This will allow for more relevant and human-like recommendations that adapt based on the user’s communication style.
    • Industry-Specific Applications: Specialized versions of the app could be developed for industries like legal, healthcare, and education, where accurate, domain-specific text recommendations could enhance productivity.

  3. Personalization & User Profiles

    • User Profiles: Over time, the app will learn and adapt to each user’s preferences, offering personalized recommendations based on their unique tone, word choice, and style. This could also extend to multilingual capabilities for a global reach.
    • Data Privacy through Blockchain: To address concerns about security, the app will use blockchain to ensure user data remains private. With decentralized control, users can opt into sharing data to improve AI performance, but retain full control over what is shared. Blockchain’s transparency will guarantee data integrity without compromising identity privacy.
    • Tailored Solutions for Different Industries: The app could offer tailored solutions for enterprises or specific sectors, with specialized tools to meet industry-specific needs.

  4. Data Security & Privacy

    • Blockchain-Enhanced Privacy: A long-term focus will be integrating more advanced security features through blockchain, addressing concerns about identity security vs. transaction security. As users interact with the app, no sensitive data will be stored unless explicitly opted in, and all communications and data will be encrypted.
    • Privacy-First Design: Following feedback from the community, I aim to explore ways to keep communication history secure, without exposing sensitive or moderately sensitive data to the blockchain. This balance between transparency and privacy will remain a key focus as the app evolves.

  5. Advanced Analytics and Insights

    • Text Analytics Dashboard: Users will have access to detailed insights into their writing, including readability scores, sentiment analysis, and suggestions for improving the impact of their text.
    • Recommendation Impact: The app could track how well recommendations perform in real-world settings (e.g., email engagement, social media interactions) to help users optimize their writing.

  6. Commercialization

    • Subscription & Enterprise Models: As the app grows, it could offer premium features for advanced personalization, industry-specific templates, or more robust analytics.
2 Likes
18

That is more appropriate now, coming to your project, the Text Recommendation app will be a must use for everyone as it tend to tackle a common challenge many people face in communication. Are you going to support multiple languages for non English speakers to be able to use your solution?

19

Am glad you found it useful buddy, learning is an evolution that never ends
While the world is evolving, only the flexible with the ability to learn evolves with it.
As obvious as it would be, during the transformation, many would still be left out lol (the rigid fellows) gear up your research and stay on top of your games buddy :+1::muscle:

2 Likes
20

Okay thank you, which means its similar and has same features like ChatGPT

1 Like