An Introduction to Opinion Mining and its Applications. Ana Valdivia Granada, 17/11/2016

Transcription

1 Sentiment Analysis An Introduction to Opinion Mining and its Applications Ana Valdivia Granada, 17/11/2016

2 About me Ana Valdivia Degree in Mathematics (UPC) MSc in Data Science (UGR) Paper about museums: Martínez de Albéniz, V. and Valdivia, A.; Measuring and Exploiting the Impact of Exhibitions Scheduling on Museum Attendance. Master Thesis about tsentiment Analysis Organizer

3 ROADMAP 1. Introduction ti 2. The Sentiment Analysis Problem 3. The Sentiment Analysis Process 4. My Master s Thesis

4 1. INTRODUCTION What is SA? Sentiment Analysis (SA) is the field of knowledge that analyses people s opinions, reviews orthoughts about products, companies or experiences identifying its sentiment. Also referred as Opinion i Mining. i

5 1. INTRODUCTION What is SA? DO NOT EVEN TRY TO VISIT - A total waste of time!!!. Spent 5 hours in the ticket queue in the broiling sun 35 degrees. An officious staff member told us when we reached the head of the queue that there were no more tickets and to buy online Alhambra with General Life parks and gardens, the tower and Nazrid palaces is absolutely amazing. If you are in Granada you must not Most missvisited it. monument in Spain. There are no words to descibe this place - beaty awaits around every corner. THe mixture of two cultures in one place makes it very special

6 1. INTRODUCTION Where it comes from? Sentiment Analysis Parsing Topic segmentation Name entity recognition (NER) Part-of-speech tagging (POS) Discourse analysis Machine translation Automatic summarization NLP

7 1. INTRODUCTION Why is SA being popular? Web 2.0 Social Networks

8 1. INTRODUCTION Customer s satisfaction analysis and applications inthe news and media industry

9 1. INTRODUCTION Why is SA being popular? Social media sentiment is the #nofilter voice of the people. analysis and applications inthe news and media industry

10 ROADMAP 1. Introduction ti 2. The Sentiment Analysis Problem 3. The Sentiment Analysis Process 4. My Master s Thesis

11 2. THE SENTIMENT ANALYSIS PROBLEM What s an opinion?

12 2. THE SENTIMENT ANALYSIS PROBLEM What s an opinion? If we cannot structure a problem, we probably bl do not understand d the problem. B. Liu

13 2. THE SENTIMENT ANALYSIS PROBLEM What s an opinion? If we cannot structure a problem, we probably bl do not understand d the problem. B. Liu

14 2. THE SENTIMENT ANALYSIS PROBLEM What s an opinion? Liu s proposal: If we cannot structure a problem, we probably bl do not understand d the problem. B. Liu. BOOK REMARK B. Liu, Sentiment analysis and opinion i mining i

15 2. THE SENTIMENT ANALYSIS PROBLEM Polarity

16 2. THE SENTIMENT ANALYSIS PROBLEM Polarity

17 2. THE SENTIMENT ANALYSIS PROBLEM Polarity

18 2. THE SENTIMENT ANALYSIS PROBLEM One example is worth a thousand words

19 2. THE SENTIMENT ANALYSIS PROBLEM One example is worth a thousand words Liu s proposal: We were very tired after a loong walk. We stopped her for a rest, the first nice thing here, is the view, and the fruit juices were excellent. We felt much better after drunk it. Also the desert were very good. Thank you.

20 2. THE SENTIMENT ANALYSIS PROBLEM Different analytic levels Document level Sentence level Aspect or entity level

21 2. THE SENTIMENT ANALYSIS PROBLEM Main concerns Different types of opinions Direct/indirect, comparative, explicit/implicit, Deal with ihtext mining i Grammar mistakes, emoticons, Irony and sarcasm Fake or spamopinions

22 ROADMAP 1. Introduction ti 2. The Sentiment Analysis Problem 3. The Sentiment Analysis Process 4. My Master s Thesis

23 3. THE SENTIMENT ANALYSIS PROCESS Step by step

24 3. THE SENTIMENT ANALYSIS PROCESS Step by step

25 3. THE SENTIMENT ANALYSIS PROCESS Sentiment identification Sentiment extraction algorithms Expert or user Stanford CoreNLP MeaningCloud s Microsoft Azure

26 3. THE SENTIMENT ANALYSIS PROCESS Step by step

27 3. THE SENTIMENT ANALYSIS PROCESS Feature Selection Bag of Words

28 3. THE SENTIMENT ANALYSIS PROCESS Feature Selection Term Document Matrix Bag of Words

29 3. THE SENTIMENT ANALYSIS PROCESS Feature Selection Term Document Matrix Bag of Words tf idf

30 3. THE SENTIMENT ANALYSIS PROCESS Feature Selection Text Preprocessing Parsing Stemming Remove STOP Words

31 3. THE SENTIMENT ANALYSIS PROCESS Feature Selection Text Preprocessing Parsing Stemming {nightmare, nighttime, nocturnal, nightlife...} night Remove STOP Words

32 3. THE SENTIMENT ANALYSIS PROCESS Feature Selection N grams More sophisticated Aspect Based Sentiment Analysis ASUM

33 3. THE SENTIMENT ANALYSIS PROCESS Step by step Medhat, Walaa, Ahmed Hassan, and Hoda Korashy. "Sentiment analysis algorithms and applications: A survey." Ain Shams Engineering Journal 5.4 (2014):

34 ROADMAP 1. Introduction ti 2. The Sentiment Analysis Problem 3. The Sentiment Analysis Process 4. My Master s Thesis

35 4. MY MASTER S THESIS

36 4. MY MASTER S THESIS Objectives 1. Study correlation between human and machine sentiment 2. Classify opinions 3.Dicover interesting patterns in negative opinions

37 4. MY MASTER S THESIS

38 4. MY MASTER S THESIS

39 4. MY MASTER S THESIS Studying correlation between different sentiment labels SentimentCoreNLP SentimentValue

40 4. MY MASTER S THESIS Studying correlation between different sentiment labels % of coincidence id

41 4. MY MASTER S THESIS Studying correlation between different sentiment labels % of coincidence id

42 4. MY MASTER S THESIS Classification problem positive positive UFSM negative BFSM negative

43 4. MY MASTER S THESIS DocumentTerm Matrix TripAdvisor Alhambra data set Use UFSM and BFSM Split it in three sets depending on sentiment class label Classification algorithms Apply different machine learning algorithms in train data set with 5cv Preprocessing If it is very unbalanced, apply oversampling techniques Split it up Split complete set in 75% training set and 25% testing set Evaluate Results Check measure values and dicuss best model

44 4. MY MASTER S THESIS XGBoost IR = 1 unigrams

45 4. MY MASTER S THESIS Subgroup Discovery negative SD Map algorithm

46 SUMMARY SA is a very challenging problem Lots of applications New research line

47 THANKS! any _ valdi