It is a search engine for searching gourmet food reviews. The data set for the reviews is taken from Amazon Fine Foods reviews which has 568,454 reviews given by 256,059 users. You can checkout the live demo here.
Clone this repo:
$ git clone https://github.com/prashkr/tasty-search.gitBefore running the following command make sure you have the dataset file (Amazon reviews) in the resources folder.
Now move into the cloned directory and run:
$ ./gradlew runMake sure you have JDK 8 installed
Your application will be up and running on the port specified in the tastysearch.yml file which is the global configuration file for our application. Access your application as follows:
http://localhost:[port]/tastysearchThe configuration has the following fields which can be changed according to needs.
- port : port on which you want to run your application
- sampleSize : number of samples to collect out of all the reviews
- indexType : type of index to use. 1 for Hash map index. 2 for Radix tree index
- resultSize : number of results to emit
- fileName : name of the data file
- testSetSize : number of queries to generate for load test
- maxTestTokens : max number of tokens in the generated query (for load test)
- testSetFileName : name of the test set file
Entries in the table correspond to the tuple (Average Latency, Throughput in requests/sec). Ran the tests for 10 minutes each.
| Structure \ Threads | 1 Thread | 10 Threads | 20 Threads |
|---|---|---|---|
| Radix Tree | (2ms, 452/sec) | (9ms, 1064/sec) | (16ms, 1177/sec) |
| Hash Map | (1ms, 470/sec) | (8ms, 1152/sec) | (15ms, 1295/sec) |
| Concurrent Hash Map | (1ms, 578/sec) | (7ms, 1317/sec) | (13ms, 1418/sec) |
| Hash Map with Heap Optimization during query | (1ms, 640/sec) | (6ms, 1403/sec) | (10ms, 1837/sec) |
| Brute Force Search | (167ms, 6/sec) | NA | NA |
MIT