Team - Lemon Potato
JianChen Zhao, HongKun Tian
With such a vast plethora of choices of restaurants within the city, each of which having hundreds if not thousands of reviews, it has become impossible for casual restaurant goers to choose optimally where to go, often leaving their dinning experience up to chance after just seeing the overall rating and reading one or two reviews.
Not only is the abundance of reviews a problem for restaurant goers, but also for restaurant owners. Should restaurants owners wish to gauge where they need to improve based on publicly posted reviews, they would need to sift through countless reviews and try to parse through each review’s content to determine which aspect requires improvement, when they could be accomplishing more productive tasks to improve their restaurant.
We wanted to develop a service to allow both restaurant goers and business owners to analyze restaurants' reviews. Our project aims to provide our users with an insightful and meaningful summary of recent reviews.
What it does
Our website provides an interactive interface to search for an existing location on Google Maps whose reviews will undergo sentiment analysis.
Our website will then display a brief summary of the contents of the reviews analyzed. This includes a one sentence summary of the reviews, and a listing of subcategory ratings.
This immediate summarized feedback is integral to allow our users to make quick and calculated decisions based on it.
How we built it
We used Google Cloud Platform (GCP) for our backend and the services GCP provides, such as App Engine, Cloud Storage, Cloud Natural Language API. Google unfortunately did not offer any form of text summarization API, so we had to leverage Aylien's Text Analysis API to summarize the retrieved reviews.
Challenges we ran into
Using the Google Maps Platform APIs proved to be far more challenging and difficult than anticipated, especially since we had no prior experience using the API.
Accomplishments that we're proud of
We somehow managed to finish our project on time. We really happy and proud of having a working project, since we were worried we might not finish.
What we learned
We learned a tremendous amount on how to work with Google Maps Platform and its APIs. We also learned on how use GCP and its services with great success.
What's next for Sureviews
There are multitudes of ways our project can be improved. As a first step, we can localize our analyser to be able to detect languages. As to accommodate not only for more regions, but also bilingual locations.
Furthermore, a more detailed analysis can be achieved by working with real world clients to access full review history. Currently, our project can only retrieve the 5 most recent reviews due to Google Maps Platform's API restrictions.
Finally, the scope of our project could be expanded to cover other types of businesses. Currently, we have chosen to limit the scope of our project to restaurants only and have pre-selected the appropriate keywords associated with the dinning industry. The versatility of our project could also be vastly improved by letting the user choose the keywords from a list of keywords they'd like to see ratings for.
We ran into some quota overlimit on appengine last minute, so the website only works locally