TwitMiner

TWITMINER 2013 RESULTS

Finally we have some good news!!! After a really long evaluation process by PhD students at IISc and ML and business teams at Amazon, we have decided to declare the following teams as the winners of TwitMiner 2013:

Rank 1: Mani Kumar Adari from RGUKT, Nuzvid.
Rank 2: Sauradyuti Coondu and Anasua Mitra from Jadavpur University, Kolkata.
Rank 3: Ahmed Shabib and Arpan Gupta from PESIT, Bangalore.


Heartiest congratulations to the winners. Amazon.com will be offering internship to Mani Kumar Adari. The second and the third ranked teams will be eligible for prizes worth 10000 INR and 6000INR respectively. All the teams who uploaded their final submissions will also be offered participation certificates!

We are extremely sorry for the delay in declaring the results! We really appreciate the patience shown by everybody. We hope you all had fun mining tweets and got interested in Machine Learning! Please watch out for a few more updates in this portal related to the results, success stories and also the codes and writeups of the best team. Hope it'll be a learning experience as well.

We are looking forward to getting your valuable feedback HERE. This will help us to make the contest BETTER and BIGGER next year.

Teams which completed the final submissions, please fill up this FORM for particpation certificates(if you haven't done already).

SUCCESS STORIES

Mani Kumar Adari(WINNER of the Amazon.com internship) about himself:
"I, A. Mani Kumar is presently pursuing B.E, Computer Science and Engineering at Rajiv Gandhi University of Knowledge Technologies, Nuzvid. A valuable step after my 10th standard into RGUKT(www.rgukt.in) made me to make a long journey to get introduce to the world of Computer Science especially Machine Learning and NLP. The great initiation of RGUKT by taking merit students from rural areas gave me this great opportunity and made me to learn Computer Science topics pragmatically through "Learning by Doing" methodology."

His Success Story:
"I came to know about Twitminer programming contest on 9th March. Without delaying any second, I started designing classification model by summing my total knowledge in NLP and ML gained in my academics. I spent my complete time till the end of contest to improve accuracy. Finally I end up with Hierarchical classification model using Fisher Score based NB and SVM which can do classification with pretty good accuracy. Thanks to Twitminer contest organizers for providing a great interface to showcase my ML potential!"

Code and writeup:
People interested in learning from Mani's work in TwitMiner can download it from HERE.

FINAL LEADERBOARD

The final scores are now available below:

Rank	Team Name	Score
1	mani2348	97.155
2	FireBird	95.748
3	HYDRA	95.656
4	Godfellas	95.564
5	Twanalysts	95.473
6	BEing	94.983
7	Turtle	94.922
8	Morpheus	94.891
9	jRh	94.433
10	Hellraisers	94.402
11	coders	94.157
12	peace	94.065
13	Robot	93.423
14	DoesNotMatter	93.331
15	viscrisn	92.077
16	AdaBoost	92.077
17	Gold_Miners	91.649
18	2_bytes	91.618
19	Numb3r	91.251
20	Golamper	90.945
21	Miner	90.578
22	Beta_Bots	90.456
23	KGP_Learners	90.364
24	Team_Jinkchak	90.272
25	Maniacs	90.242
26	Bazinga	90.211
27	codeinsleep	90.150
28	Banyan_Learners	89.997
29	Stackoverflow	89.966
30	try_catch	89.936
31	heritage	89.936
32	Dark_Horse	89.813
33	Matrix	89.599
34	KGP_tweets	89.538
35	MAC_LEARN@IITKGP	89.507
36	Deagle	89.507
37	MS1	89.049
38	tweet#01	88.957
39	fundoers	88.682
40	42	88.682
41	skyfall	88.651
42	Pegasus	88.590
43	Twister	88.498
44	Tihana	88.131
45	Extractorz	88.009
46	The_Anonyminers	87.764
47	AustereSapiens	87.244
48	Cyber_Punk	86.938
49	plutos_pirate	86.754
50	Zen_Coders	86.693
51	Metal_Striker	86.693
52	firefly	86.693
53	BinaryMiners	86.632
54	kgpMiners	86.173
55	TheTeam	86.020
56	Useless_Productions	85.959
57	Miners1	85.959
58	W-TransposePhi	85.806
59	Lazy_Learners	85.500
60	Inbetweeners	85.347
61	W00t	85.102
62	Vamshi	84.980
63	TextWreckers	84.644
64	TwitMiners	84.154
65	Vetri	84.001
66	TwMiners	83.940
67	Twitterati	83.665
68	Infominers	83.114
69	Zoid	82.686
70	DH@IITKGP	82.533
71	RocketShip	82.319
72	MachineTeachers	81.034
73	LetsTry	80.942
74	Arjunvmm	80.697
75	Prometheus	74.518
76	Sam	73.876
77	RoHaN	73.509
78	The_Unsinkables	72.958
79	Specialitas	68.553
80	kungfupanda	63.261
81	Underdogs	61.578
82	tweet-o-hola	54.787
83	minesweber	49.648
84	StupidAmigos	0.489
85	CASL	0.336

FINAL SUBMISSIONS(CLOSED)

As a part of the final submission to this contest you are required to submit the following:

1. Predictions on the test data in the same file format as the validation data; click here for a sample file (in .txt format).
2. A brief writeup of your algorithm in the format prescribed here (in .pdf format).
3. Properly documented code along with a README file containing instructions to run the code (archived as a .zip).

Final submissions can be done HERE after login and will be allowed till March 15th 11:59PM IST (GMT + 5:30hrs). Multiple submissions are allowed but only the latest will be considered.

Final Submissions are CLOSED. Please mail us if you could not submit. Thanks a lot for participating!

TEST DATASET

Download Link: Test data

Please read the Instructions.txt file carefully. Also note that there might be noisy tweets included in the test dataset. But while calculating scores we wont consider them. The scores for the test data will not be displayed during submission. They will be displayed after the contest closes.

---

VALIDATION LEADERBOARD

Click here to access the upload form and the leaderboard.

While uploading your prediction labels on the validation data in a text file (.txt), use the following format:
< tweetid label >
where the 'label' is either 'Sports' or 'Politics' assigned to the tweet with identifier 'tweetid' by your algorithm. Note that the prediction for each tweet must be in a separate line, with a single blackspace separating the tweetid and label. (Please be careful to not include additional blank spaces in the submitted file.)

Here is a sample file that you can look up for the format.

TRAINING DATASET

Download Link: Training and Validation data
Please read the Instructions.txt file carefully.

---

REGISTRATION

Registrations have been closed.