Analyzing sentiment is difficult. You need a ton of data to get a real picture about your users' thoughts and opinions but computers are notoriously bad at parsing nuance in at parsing nuance in language. Sentiment analysis gets even more difficult when you’re looking at social media, where messages are rife with sarcasm, misspellings, acronyms, and images.
The problem is, most companies use out-of-the box solutions like natural language processors. Those tools top out at about 70% accuracy. And those tools really can't go much deeper than "positive" and "negative."
Join this webinar to listen to Randall Sparks, Pallika Kanani (Oracle) and Lukas Biewald (CrowdFlower) talk about how tapping into a global on-demand workforce helps increase accuracy of these sentiment mappings and results.
Randall Sparks is Principal Member of Technical Staff at Oracle Data Cloud in the social media analytics group. Randall has a Ph.D. in Linguistics from Univ. of Colorado. He has over 25 years experience in in software development, mostly in NLP, text analysis, media, telecommunications services. Randal is an inventor with 5+ patents filed and granted.
Pallika Kanani is Senior Research Staff Member at Oracle Labs. Pallika has a Ph.D. in Computer Science from Univ. of Massachusetts Amherst and an M.S. in Computer Science from NYU. She works with the Information Retrieval and Machine Learning Group at Oracle Labs. Current projects include: Working on large scale information extraction and ranking problems in various business domains.
Lukas is the CEO at CrowdFlower. Following his graduation from Stanford University with a B.S. in Mathematics and an M.S. in Computer Science, Lukas led the Search Relevance Team for Yahoo! Japan. He then worked as a senior data scientist on the Ranking and Management Team at Powerset, Inc., acquired by Microsoft in 2008.
People talking about your brand is one thing. Understanding what they're saying is a different story altogether. CrowdFlower's people-powered sentiment succeeds where automated solutions like NLP fail. Here's how: