Geosocial networking app
The application represents a geosocial networking app that uses geolocation and the user’s time to offer relevant recommendations for nearby locations, places of interest, and events.
Geosocial networking apps are a relatively new type of social apps. The most popular players on the market are Yelp, Facebook Places, and Foursquare. These apps allow users to share their locations as well as find recommendations for locations, or ‘venues’.
Sharing our location on online social networks has great advantages: it can help us find our way, pick restaurants and shops, and even locate nearby friends and other people within the selected radius. Our client wanted to create a unique geosocial networking solution powered by AI for the local market.
The required recommendation system has to use:
- Computer Vision to analyze images;
- NLP technology to analyze chat and image captions, and automatically pinpoint topics of interest, locations, dates and time for proper advice;
- Recognition of voice commands in several languages; support of topic modeling of social messages, as well as syntactic text simplification of complex sentences in the information repository structured as a graph.
- May 2021 - Ongoing
- Seattle, WA, USA
- Media & Entertainment
- Amazon AWS
- REST API
- Google Maps
- Amazon S3
- Amazon EC2
- Amazon SQS
- NLTK gensim
- Calendar flow The information in the app is organized according to the calendar flow, formed based on the time and location of the user
- Topic modeling The content is broadcast from the location where the user is connected. It’s also broken down into thematic channels: news, food, housing, local stories, etc
- In-app chat Users can communicate directly in the app, and attach photos and videos to their messages. Post commenting is also available
- Recommendation system Based on information from the chats, images, and comments, the system offers users customized recommendations. For example, if a user is going to Kyiv for business, the system will offer relevant Kyiv-based locations so the user can plan his trip accordingly
- Bilingual recommendation system support The system supports two languages, English and Vietnamese. Thanks to voice and language recognition technology, it’s now possible to avoid typing commands or requests
- Voice assistant We used an out-of-the-box solution for the development of Voice assistant, Voice navigation, and Voice translation
Geosocial networking apps are a relatively new type of social apps. The most popular players on the market are Yelp, Facebook Places, and Foursquare. These apps allow users to share their locations as well as find recommendations for locations or ‘venues’.
Our client wanted to create a unique geosocial networking solution powered by AI for the local market.
- The client’s main idea was to change a commonly-used approach to social media posting and social networking apps, add voice and language recognition and create a unique recommendation system;
- Our client was looking for a development team with experience in Machine Learning;
- Our client wanted to start as fast as possible and create a working prototype of a recommendation model for his investors.
Our team was in charge of the project, starting with the discovery phase and POC, and finishing with development of the app prototype.
- Our Business Analysts conducted profound research of the geosocial apps market, defining competition and the most popular features.
- We developed a customizable recommendation system, based on users’ geolocation data in two languages: English and Vietnamese.
- We made our solution bilingual for English and Vietnamese-speaking users.
- We created a knowledge graph, with a lot of valuable user’s data for further analysis.
- We also added syntactic text simplification of complex English sentences in the graph for developers convenience.
The geosocial networking project is very interesting and technically challenging, and that’s the greatest appeal of it! We have tested a lot of models and technologies to solve certain problems. The most difficult part of the project, probably, was that at the start of the project, we did not have enough real data. The most memorable thing for me was the creation of a knowledge graph for the social network. The task turned out to be very multifaceted and included many other subtasks.