Munchee Brings Blockchain to Food Reviews and Delivery – ICO Soonbr>
Mobile food review app Munchee, which uses reviews and recommendations of individual restaurant dishes via photo and videos, is creating MuncheePlatform, a blockchain social platform that incentivizes users and restaurants. The platform looks to disrupt traditional restaurant review sites such as Yelp, FourSquare, Google Places, and Zagat by introducing a blockchain user review process that is based around a crypto-token to incentivize ecosystem participants.
Users will receive MUN tokens for actions within the application, including approved, peer-reviewed food recommendations. Tokens can then be publicly traded or can be redeemed for special offers from restaurants and, eventually, from food delivery services. To incentive user to hold their tokens, the company will introduce a tiered rewards program for users that will be based upon the amount of MUN tokens a user is holding on the ethereum blockchain. The higher the tier, the more rewards a user will receive for actions on the platform.
Block Tribune talked with Dr. Sanjeev Verma, co-founder and CTO of Munchee.
BLOCK TRIBUNE: What was the basis for starting this company? Did you have a bad experience with Yelp or Google?
SANJEEV VERMA: As all other food review apps are not focused 100% on reviewing restaurant dishes, it takes foodies so much time to be able to find that one good dish, as reviews are usually about the restaurant instead of specific dishes. Non-food factors can also negatively affect a restaurant’s overall image. With Munchee, we believe that even grade C restaurants do have a couple of grade A dishes. With current centralized platforms operated by Yelp, FourSquare, etc., it is impossible to prove that there is truth in the review and rating system. In fact, Yelp has long been accused of review manipulation for profit purposes. Munchee’s goal is to be the trusted platform that provides high-quality reviews with our peer-reviewed system, which users are able to verify the validity of others’ posts without Munchee’s interference.
BLOCK TRIBUNE: Tell us about the incentives to post.
SANJEEV VERMA: Due to the lack of incentive for the standard end-users, those who have neutral or positive experiences are far less likely to post reviews than those who have negative experiences. With that being said, Munchee will be rewarding users with MUN tokens for their participation on the platform, which can be used to exchange for food and services at our partner restaurants. Consumers and business owners are charged exorbitant amounts in terms of interest rates, fees and surcharge for their services by existing centralized financial networks such as VISA, Mastercard etc. With the blockchain technology, Munchee will be eliminating the need for restaurants to pay outrageous transaction fees to third-party payment networks.
BLOCK TRIBUNE: How do peer reviews create fair play?
SANJEEV VERMA: Munchee’s Peer Review system is implemented to ensure that only quality reviews are accepted in the Munchee ecosystem. Every new submission first goes through the peer review by other Munchee users, which makes the Munchee platform truly social and transparent. Munchee uses Machine Learning to mine user preferences to suggest suitable reviewers for peer review based on criteria, such as location, language, food category, food preference, etc. Reviewers also need to have a good history on the platform to be able to review others’ posts.
Everyone in the Munchee ecosystem participates in improving the quality of the platform through active participation–instead of being just passive user of the content provided by others. This fosters valuable content and a collaborative community. Such a platform attracts many first-time users to the platform since they can’t just only access trusted content but also earn MUN tokens by actively participating in improving the validity and quality of the content.
BLOCK TRIBUNE: What constitutes a “low-quality submission” in your ecosystem?
SANJEEV VERMA: A post submitted to Munchee will be evaluated based on two criteria: Validity- the existence of the restaurant, the accuracy of the restaurant address, the existence of the reviewed dish in the menu, the photo, etc. This is an onjective peer-reviewed process. Quality- the support from the community portrayed by the number of likes and pins from other users for the post. These are evaluated based on the quality of the posted photo and the review itself. These reviews must have passed the peer-reviewed process prior to being evaluated for quality.
BLOCK TRIBUNE: How will you prevent a community of trolls from developing?
SANJEEV VERMA: We use machine learning to filter out the reviews that contain spammy information first so users won’t need to review spammy submissions. Since we rely on the community, we can argue that if a user posts a lot of spam reviews, the community will know, one person may not know exactly, but 10 people may know better. Then if a review is marked as a bad review for the first time, we will reduce the rewards that he will get next time, if next time it’s again marked as a bad review, we will reduce more, the third time we will ban him from the system.
BLOCK TRIBUNE: How will you deal with libel and slander issues?
SANJEEV VERMA: Munchee will not publish every review. Majority of the reviewers will have to agree to a certain peer-review before the review will be accepted in the platform. Our machine algorithm will also look for certain “keywords” to flag any such review. In more detail, for example, if someone hates a restaurant and he submits a bad review for a dish of that restaurant but he has never eaten that dish in real, just take some pictures and put a bad review as a slander. We can do 3 things about this: We can check approximately if that dish belongs to a restaurant by matching the dish with the menu of the restaurant. This approach may encounter a problem of synchronization, means that sometimes the restaurant owner is too lazy to update the menu to our system. By using some heuristic, a Western restaurant is not likely to serve the Peking roasted duck. I use the word “likely” here to emphasize that we cannot know exactly, I know some cases that a Western restaurant does server some Chinese food, but this way is a good heuristic and it should be correct in most cases.
By trusting the community, we can argue that if a review contains libel or slander, the community will know, so they will rate the review badly. We can combine 1) and 2) in the Machine Learning (ML) system to flag the review that the ML system suspects that it’s a libel or a slander. If a review is flagged, we increase more reviewers to review it and we use 3) to argue that more reviewers means that the chance to detect that this review is libel is higher. This follows the probability theorem, for example, you toss the coin 2 times and you get 1 head and 1 tail and you claim that the probability to get the head of the coin is 0.5, but does this sound correct? If you toss the coin 10 times and you get 5 heads and 5 tails, it’s likely that the probability of to get head of this coin is truly 0.5. This case is the same, for example, in normal review, we pick 5 reviewers, but for flagged review, we pick 20 reviewers, it’s likely that more reviewers will produce better judgement and evaluation for a review.
BLOCK TRIBUNE: Will businesses have the right to appeal any reviews they deem unfair? If so, how will that be reviewed?
SANJEEV VERMA: Munchee will be using peer review rating system and machine learning techniques to prevent fake reviews from appearing on our site. We will also be keeping chronological records of all reviews appearing on our site by storing fingerprints of the reviews on the Blockchain. Each fingerprint will carry digital identity of the reviewer (public key), digital signature of the review and a time-stamp. This basically means that we can always link the review to a certain reviewer.
We will encourage businesses to appeal any review so that we can do our investigation. We will also solicit the feedback from our reviewers with five star ratings to ensure that the system is fair to the businesses. However, if the reviewer is found to be genuine and a large number of Munchee users agree to the review in dispute then businesses will also have to agree with the review. Munchee wants to implement a fair and decentralized restaurant rating system and it will be in our interest to ensure that the system is fair.
BLOCK TRIBUNE: How will your site make money?
SANJEEV VERMA: Ad revenue from restaurants and commission from food delivery partners
BLOCK TRIBUNE: Tell us about your ICO plans.
SANJEEV VERMA: Yes, we are planning an token sale event. Our pre-sale event starts on Oct 31 and will run for 2 weeks before the toke sale event begins on Nov 14 and will run for a period of 30 days. A 20% discount during our presale is offered for early token buyers. Token issuance and distribution will be announced shortly.