2016, 10 trends in text analytics, sentiment analysis, and social analytics

Text analytics, sentiment analysis, and social analytics help you transform the "voice" of customers, patients, the public, and the market on a certain scale. The technology is now widely used in a range of industrial products, from healthcare to finance, media, and even customer markets. They extract business insights from online, social networks, and enterprise data sources.

It's a useful thing to extract insights from text, audio, images, and network Connections!

The current analysis technology is still quite good, although in some areas, such as digital analysis and market research are slightly behind. But even in terms of "customer experience, community listening, user interaction," there is still a lot of room for development. This fast-growing market space means plenty of opportunities for new players and long-time veterans.

As technology and applications continue to converge, it is better to observe the overall effect than to independently examine each area of analysis. Social analysis that ignores emotions is incomplete, and we really need text analytics to get social sentiment data and investigate emotional data from the Web.

In the coming 2016, a forward-looking observation is made of the development trend of text analysis, affective analysis and social analysis.

One, multi-language is the kingly

While plain English text analysis has remained normal, it is much better to do just one language well than to include many languages. Machine learning and MT have taken a big step towards multilingual text analysis, making it a new standard. But if you do need to develop more languages, do some research in advance: Many developers are strong in their core language, but weak in other languages. So when you choose, be careful.

Second, the text analysis obtains the recognition

Text analytics capabilities are a key solution for customer experience, market research, customer insight, digital analytics and even media reviews, and each text analytics service provider is constantly competing for the strengths of analytics capabilities. The general trend is "quantitative qualitative", and it is important that text analysis be incorporated into business solutions.

Three, machine learning, statistics and language engineering coexist

Tomorrow belongs to machine learning, recurrent neural networks and similar technologies, but today, the long-established language engineering approach still prevails. I'm referring to the classification system, the parser, the lexical and syntactic networks, and the system of sentence rules. At present, we are in a "blossoming, the schools of contention" era, so many ways can coexist. For example, even the leaders of crowdsourcing data processing: Crowdflower all embrace machine learning, and startups Idibon combine tradition and modernity as a big selling point: "You can build a custom classification system and tweak them using machine learning, rules, and dictionaries/patterns that you already have." ”

Iv. image analysis into the mainstream

The world's leading image analytics provider has applied image analysis technology to the brand signal interpretation of social media---don't believe you look at pulsar and crimson Hexagon---and through machine learning, image analysis technology has become IBM's acquisition of Alchemy in 2015. A major selling point of the API. Indeed, Metamind, a hot start-up, transformed from NLP to image analysis in 2015, due to the enormous opportunities behind its image analysis.

Five, a breakthrough in speech analysis, video analysis followed by

The entire market likes to talk about multichannel analytics and user journeys, which involve multiple contacts. and social networks and web media are flooded with video, words, and non-textual forms of language elements, including intonation, speed, volume and repetition, all convey meaning, and these meanings can be obtained through speech analysis and voice to text. Not just the customer service center, in 2016, all market researchers, publishers, research and insight professionals are constantly looking for breakthroughs. It can be expected that future speech analysis will also become an important force to promote the development of MMI interface.

Vi. Extended sentiment analysis

Advertisers have long recognized that emotions can change consumers ' decisions, but until recently, extensive and systematic research into emotion and decision-making has transcended our capabilities. According to your perspective, enter into affective analysis, either as a subclass of affective analysis, or as a sister class. With the purpose of quantifying our emotional responses, we use facial expression analysis to extract our emotional state from images and videos (or from speech or text). Service providers in this area are: Affectiva, emotient and Realeyes for video services, beyond verbal for voice services, and Kanjoya for text services, including advertisers, media, market researchers and agents.

Seven, ISO network expression analysis

We already have text, images, voice, video, and so on, so why should we use Web expressions? Because they are concise, easy-to-use, vivid and interesting, they complement and impact on long-form content, which is why internet slang is extinct. Facebook is trying to dig up web emoticons, and better yet, we've seen variants like line stickers. Now all we need is the network expression analysis. Technology in this field is rising through startups like Emogi. While most people use counting and sorting to get network emoticons, like Instagram engineer Thomas Dimson and Slovenian research organization Clarin.si do. But some of them, such as SwiftKey, are worth paying attention to.

Eight, network + content depth of insight

This is both my forecast for the 2016 trend and I also mentioned in a 2015 interview with data scientist Preriit Souda of TNS, a market research firm. Preriit points out: "The network gives the structure to the session, and content mining gives meaning to it." "Insight comes from the understanding of information and connections, and also from how connections are activated." So add a graphical database and Web visualizer to your toolkit, which is why Neo4j.js and Gephi are so successful. Setting up a data analysis platform similar to Qlikview is also a choice, a choice that can be combined with text and digital analysis, which is a must for 2016 years.

Nine, 2016, you'll read (or interact with) much more machine-written content

The technology for machine writing is called Natural language synthesis (Natural Language GENERATION,NLG), which provides the ability to compose articles, letters, short messages, abstracts, and translations from text, data, rules, and content based on algorithms. NLG is for high-volume, repetitive content: finance, sports, and weather forecasts. The relevant service providers are Arria, narrative science, automated Insights, data2content and Yseop. You can also look at the machine side of your conversation with your beloved virtual assistant: Siri, Google now, Cortana or Amazon Alexa, or automated customer service, other programmatic systems. These systems are all categorized in natural language interactions (Natural Language Interaction,nli), where artificial solutions is worth a look.

Ten, machine translation gradually mature

For a long time, people have been hoping for a universal translator like Star Trek, but since 1950 scientists have said that machine translation can be achieved within 3-5 years, accurate, trustworthy machine translation has been a mystery. (ACM queue Author "Natural Language translation at the intersection of AI and HCI" fully discusses the status of machine translation under the condition of human-machine integration) I can't say victory is in sight, But thanks to big data and machine learning, 2016 (or 2017) for most tasks, the mainstream language machine translation can be good enough. This is victory!

Summarize

If you are a text analysis, sentiment analysis, or social analyst, solution Provider or user, every trend will affect you, both directly and indirectly. Because human data has now been woven into the web of technology that we live on every day. The line connecting this net is more data, more efficient use, to create a life-changing machine intelligence.

Original: Ten text, sentiment, and social analytics trends for 2016

2016, 10 trends in text analytics, sentiment analysis, and social analytics