CS297 Proposal

AI Dining Suggestion App

Bao Pham (bao.t.pham@sjsu.edu)

Advisor: Dr. Chris Pollett

Description:

Trying to decide what to eat sometimes can be challenging and time-consuming to some people. Given the main sources of data come from Google and Yelp APIs along with other factors that influence food decisions such as time, price range, traffic, temperature, etc., the goal is to build an AI model that can learn from one's dining pattern over time to help make restaurant suggestions (with maybe some interesting options) at any time.

Schedule:

Week 1: (9/4-9/11)	Kickoff meeting with Professor Pollett and start defining requirements
Week 2: (9/11-9/18)	Start building API from Google [6] and Yelp [7] with restaurant data
Week 3: (9/18-9/25)	Continue on building API with simple identification system to set up user data
Week 4: (9/25-10/2)	Polish and test API with modifications (Deliverable 1 due: 10/2)
Week 5: (10/2-10/9)	Read and discuss with Professor Pollett about extra that need to be collected as well as next steps for API
Week 6: (10/9-10/16)	Start implementing extra services to collect more data
Week 7: (10/16-10/23)	Polish and test extra services(Deliverable 2 due: 10/23)
Week 8: (10/23-10/30)	Discuss with Professor Pollett about UI implementation
Week 9: (10/30-11/6)	Start designing and building new UI
Week 10: (11/6-11/13)	Implement features for front-end
Week 11: (11/13-11/20)	Polish and test with (possibly) real users to collect data. Observe and make modifications to front-end UI (Deliverable 3 due: 11/20)
Week 12: (11/20-11/27)	Start reading papers for possible AI model architecture [1]-[5]
Week 13: (11/27-12/4)	Gather data and start building AI model using Tensorflow
Week 14: (12/4-12/11)	Train and observe model (Deliverable 4 due: 12/11)
Week 15: (12/11-12/18)	CS 297 report (Deliverable 5 due: 12/18)

Deliverables:

The full project will be done when CS298 is completed. The following will be done by the end of CS297:

1. Create an API to query and process restaurant related data from Google and Yelp given a specific GPS location such as names, price levels, food types, images, etc. (possibly with Facebook authentication system for users if time permits).

2. Implement additional services to get extra information for other factors that influence food decisions (temperature, traffic, ambient, etc.) that are not provided from Google and Yelp APIs

3. Build simple UI to start collecting data to train the AI model with different users. This might need other services to get realistic data.

4. Start building and training data once enough data is collected for the first round. Justify as well as improve the AI model after training.

5. CS 297 report.

References:

Shuai Zhang, Lina Yao, Aixin Sun: Deep Learning based Recommender System: A Survey and New Perspectives, 2017.
Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, Rohan Anil, Zakaria Haque, Lichan Hong, Vihan Jain, Xiaobing Liu: Wide and Deep Learning for Recommender Systems, 2016.
Covington, Paul, et al. Deep Neural Networks for YouTube Recommendations Google AI. Google AI, 1 Jan. 1970, ai.google/research/pubs/pub45530.
Representation Learning of Users and Items for Review ... pdfs.semanticscholar.org/4946/89f4522619b887e515aea2b205490b0eb5cd.pdf.
Marina Kobayashi, Victoria Sosik, David Huffaker. Not Some Trumped Up Beef: Assessing Credibility of Online Restaurant Reviews. 15th Human-Computer Interaction (INTERACT), Sep 2015, Bamberg, Germany. Lecture Notes in Computer Science, LNCS-9298 (Part III), pp.116-131, 2015, Human-Computer Interaction INTERACT 2015.
https://developers.google.com/places/web-service/intro
https://www.yelp.com/developers/documentation/v3/business