CS298 Proposal
Recipe Suggestion Tool
Sakuntala P Gangaraju (priya.gangaraju@gmail.com)
Advisor: Dr. Chris Pollett
Committee Members: Dr. Jon Pearce & Dr.T.Y.Lin.
Abstract:
There is currently a great need for a domain specific search engine to search cooking recipes based on ingredients. Current search engines do not provide this feature. Many recipe search engines are available (like www.recipepuppy.com, www.supercook.com, www.cooks.com). As per my analysis these search engines do not provide / need further improvement in the following areas:
- Recipe search by ingredients.
- Better clustering of data returned by search queries.
- Provide search filters (like searching within the recipes submitted in the last X number of days).
- Provide links to view images of ingredients of a recipe.
- Provide links to grocery stores carrying the items listed in the recipe based on the location of the user.
- Suggest recipes to users based on their previous searches.
- Provide option to modify the quantities of ingredients based on servings selected by users.
My project aims to combine the features available in different search engines under one search engine and provide an intuitive interface including
the new features listed above.I plan to explore different distance measures that can be used to cluster recipe data matching user’s queries. As part of this project I also plan to build custom search engine wrappers around existing search engines to help search images of ingredients and provide local listings of grocery stores carrying ingredients listed in a recipe.
CS297 Results
- Studied Dr.Pollett’s Search Engine code.
- Implemented K-means Clustering algorithm.
- Implemented Clustering through Minimum Spanning tree using Kruskal's algorithm.
- Implemented Custom Search Engine Wrapper for Image based search.
Proposed Schedule
| Week 1 & 2:Aug.25 - sep 7 | Modify crawler to make it focused and build the recipe database. |
| Week 3:Sep.8 - 14 | Explore :
1. different distance measures used for clustering recipes.
2.different ways to identify cluster centroids. |
| Week 4 & 5:Sep.15 - 28 | Provide search filters like suggesting recipes to users based on their previous searches and searching recipes submitted recently. |
| Week 6 & 7:Sep.29 - Oct.12 | Provide a mechanism to search recipes based on ingredients and suggest alternate ingredients for user. |
| Week 8 & 9:Oct.13 - 26 | Develop the web based user interface to include features like registering, saving recipes to favorites, suggesting recipes based on previous searches. |
| Week 10 & 11:Oct.27 - Nov.9 | Modify custom search engine wrapper to include image based and location based search. |
| Week 12 & 13 :Nov.10 - 23 | Bug fixing. |
| Week 14:Nov.24 - 30 | Prepare CS298 final report. |
| Week 15:Dec.1 - 7 | Complete draft report for committee review; prepare project presentation slides. |
| Week 16:Dec.8 - 14 | Defense in front of committee. |
Key Deliverables:
- Software
- Focused crawler to build the database
- Modified clustering algorithm which includes features like suggesting recipes based on user's previous seaches and searching based on ingredients.
- Web based user interface
- Modified custom search engine wrapper to include image based and location based search
- Report
- CS298 Project Report
- Code Documentation
Innovations and Challenges
- Building up the database of recipes will be challenging.
- Defining the distance measure to cluster the recipes will be challenging.
- Providing search by ingredients and suggesting alternate ingredients will be challenging
References:
[Wang2008]Substructure Similarity Measurement in Chinese Recipes. Wang L., Li Q., Li N., Dong G., Yang Y. www 2008. 2008
Google Ajax APIs retrieved from http://code.google.com/apis/ajax/ |