Ten trends in social media analytics in 2016
A good article worthy of translation for review of what to expect from the near future. Small inaccuracies that will be noticeable to experts have no effect on the strong base of trends and trends of the development of a new “social society” gathered in one place, the features of which are more clearly manifested in reality: chat bots, emoji, and self-destructing information, shift info -use and info-generation in the direction of video, linguistic processing and identify facts in huge amounts of unstructured content ... So we strongly advise reading. Read and compare with your own vision of the development of the world around you.
A brief conclusion : "The general trend is more and more data, they are used more efficiently - to create smart" automation ", which will form the information image of the future."
[Picture from Sostav.ru]
=============
Analysis of texts and moods in social media helps to keep abreast of the views of the buyer, patient, public or even the whole industry. Such technologies are already used to the full in a wide range of different disciplines, from healthcare to finance, media and consumer market. Technology helps extract important facts and numbers from the general flow of data.
')
At present, analytics has risen to a rather serious level, although in some areas, such as digital analytics and market research, it is slightly behind. But even in the areas of its active use, such as customer support and “listening” to society and interaction with the public, there is room for improvement, and this applies both to new technologies and to an increase in the amount of information analyzed. And in a similar, still developing area there is a place for both new members and already recognized leaders.
You can consider each area of analytics separately, but where it is better to consider them all at once, as they use similar tools and technologies. Social analytics that do not take into account the tonality of the statement is incomplete, and to accurately determine the tonality of the statements, you will definitely need text analytics. In this article we will try to look into the near future and highlight the key ideas of the year, highlighting 10 trends in the analysis of texts, moods, as well as social analytics, which will gain weight in 2016.
1. Multilingual - usually
Though there are only analytics tools that support only English, guided by the principle that it is better to operate with one language rather than many at once, but the already existing Machine Learning (MO) tools allow to implement multilingual analytics, making it a new universal standard. But if you just need analysis of information in several languages, then do not lose your vigilance: many analytics providers are strong in their main language, and in the rest - not really. Choose wisely.
2. Content analytics will be universally recognized.
The use of content analytics is the key to high-quality user support, market research, opinion gathering, digital analytics and media media parameterization, and providers of such services actively compete with each other in the quality of analysis provided. Creating your own system or subscribing to an existing one is your choice, and both options have their advantages. And although such a trend can be called quantitative quality, the fact that text analysis is finally recognized as a complete solution for business is really significant.
3. Interaction of machine learning, statistics and language engineering
The future belongs to an in-depth analysis, recurrent neural networks and the like, but in the present, already established methods of language engineering dominate. There are taxonomy, parsers, lexical and semantic networks. Thus, we get a market in which “ hundreds of flowers bloom and hundreds of schools compete ... ” and all these approaches can coexist peacefully with each other. And even a company like CrowdFlower adopts machine learning methods, and Idibon attracts customers by combining classical techniques and the latest innovations: “ You can create your own taxonomies and use them in conjunction with the principles of machine learning and existing databases and dictionaries and”.
4. Image analysis is entering the mainstream.
Leading analyst vendors are already using image analysis technologies - for example, Pulsar and Crimson Hexagon , and the ability to analyze images using in-depth analysis was one of the key factors in acquiring IBM AlchemyAPI . A new, promising startup MetaMind , launched in 2015, states image analysis as its main feature, as it sees great prospects and opportunities in this.
5. Breakthrough in automatic speech and video analysis
“Vsekanalnaya” analytics, coupled with the complete decision-making process about buying a consumer - a favorite topic of the market. And social networks, where people most often express their opinions, are filled to the brim with video recordings. Pronounced words and non-textual elements of speech, such as intonation, speed of pronunciation, loudness and repeatability - have a certain meaning available for means of speech analysis and translation into text. We assume that in 2016, the use of such funds will significantly expand and they will begin to be used by marketers, editors and market research specialists. Also, speech analysis will most likely be used to improve the quality of computer interaction interfaces (including chat bots).
6. Advanced Emotion Analysis
Advertising specialists have long understood that when a person buys emotions, they are driven, but until recently a systematic study of reactions was out of technological reach. Depending on your perspective, start using emotion analysis or tonality analysis. The emotional state is established on the basis of an image using analysis of features and facial expressions (or from speech or text), with the goal of structuring the person's reaction to what he sees, hears or reads. Similar services for video recordings are already provided by Affectiva, Emotient, Realeyes, Beyond Verbal for speech, and Kanjoya for text; The number of users of such tools is growing rapidly and is used by many agencies, marketers, media and advertising specialists.
7. Analysis of emoji (emoticons)
We are subject to a huge number of information channels - text, images, speech, videos and likes. Why then use Emoji? Yes, because they are cool and expressive! Like # hashtags , they are more extensive forms of content. That's why Internet slang is almost dead (ROFL!) And Facebook is experimenting with reactions to emoji, and various alternatives appear, such as Line stickers . It is quite obvious that the analyst Emoji becomes absolutely necessary. The right technologies are already being used by various startups like Emogi . And although most of the projects do not go beyond the calculation and classification of Emoji, for example, Instagram engineer Thomas Dimson and research organization CLARIN.SI are engaged in such analysis, and some of them, like SwiftKey , are definitely worthy of attention.
8. Greater information retrieval from web content
With these words, we can characterize the general information trends of 2016, and it was this title that I gave an interview with Praitt Suud, a specialist in studying data in TNS . Praith notes that, “The web gives the dialogue a structure, and extracting the content gives it meaning .” Useful information is obtained by understanding the content and relationships with previously obtained information, as well as understanding the mechanism of the appearance of these relationships. So add to your toolkit a means of visualizing network content, because that's why companies like Neo4j, js, and Gephi (and that's not all) are so successful. Using a data analysis platform, such as QlikView, is an option that can be used with digital and text analytics tools: another item on the to-do list for 2016.
9. In 2016, you will consume much more automatically generated content.
The technology for automatic content generation is called Natural Language Generation (NLG) and allows you to write articles, e-mails, text messages and translations automatically, based on the analyzed text, grammar rules and context. NLG is the best solution for frequent, repetitive content, such as sports, finance, and weather reports, for example. These services are provided by Arria, Narrative Science, Automated Insights, Data2Content, and Yseop . You can also use the “services” of your favorite virtual assistant - Siri, Google Now, Cortana, or Amazon Alexa , or an automatic customer service system. Such systems fall into the Natural-Language Interaction (NLI) category, and services like Artificial Solutions will definitely be useful.
10. Machine translation will mature and settle down.
We have long been dreaming of a universal interlanguage translator (as in Star Trek films), but at least researchers in the 50s of the last century stated that the problem of machine translation would be solved within 3-5 years, but high-quality machine translation turned out to be the goal where more difficult goal than they thought. We will not argue that the full resolution of the issue is already on the horizon, but thanks to Big Data and machine analysis, 2016 (or 2017) will become that year when machine translation from the most common languages of the world will finally be good enough for most tasks. And it pleases!
Each of these trends will certainly affect us in one way or another, will directly affect - if you are analyzing text, tonality or social analytics, you consume or provide technical means - or indirectly, since the analysis of human data is already closely interwoven into the fabric of the information universe.
The general trend - more data, they are used more efficiently - to create smart “automation”, which will form the information image of the future.
A brief conclusion : "The general trend is more and more data, they are used more efficiently - to create smart" automation ", which will form the information image of the future."
[Picture from Sostav.ru]
=============
Analysis of texts and moods in social media helps to keep abreast of the views of the buyer, patient, public or even the whole industry. Such technologies are already used to the full in a wide range of different disciplines, from healthcare to finance, media and consumer market. Technology helps extract important facts and numbers from the general flow of data.
')
At present, analytics has risen to a rather serious level, although in some areas, such as digital analytics and market research, it is slightly behind. But even in the areas of its active use, such as customer support and “listening” to society and interaction with the public, there is room for improvement, and this applies both to new technologies and to an increase in the amount of information analyzed. And in a similar, still developing area there is a place for both new members and already recognized leaders.
You can consider each area of analytics separately, but where it is better to consider them all at once, as they use similar tools and technologies. Social analytics that do not take into account the tonality of the statement is incomplete, and to accurately determine the tonality of the statements, you will definitely need text analytics. In this article we will try to look into the near future and highlight the key ideas of the year, highlighting 10 trends in the analysis of texts, moods, as well as social analytics, which will gain weight in 2016.
1. Multilingual - usually
Though there are only analytics tools that support only English, guided by the principle that it is better to operate with one language rather than many at once, but the already existing Machine Learning (MO) tools allow to implement multilingual analytics, making it a new universal standard. But if you just need analysis of information in several languages, then do not lose your vigilance: many analytics providers are strong in their main language, and in the rest - not really. Choose wisely.
2. Content analytics will be universally recognized.
The use of content analytics is the key to high-quality user support, market research, opinion gathering, digital analytics and media media parameterization, and providers of such services actively compete with each other in the quality of analysis provided. Creating your own system or subscribing to an existing one is your choice, and both options have their advantages. And although such a trend can be called quantitative quality, the fact that text analysis is finally recognized as a complete solution for business is really significant.
3. Interaction of machine learning, statistics and language engineering
The future belongs to an in-depth analysis, recurrent neural networks and the like, but in the present, already established methods of language engineering dominate. There are taxonomy, parsers, lexical and semantic networks. Thus, we get a market in which “ hundreds of flowers bloom and hundreds of schools compete ... ” and all these approaches can coexist peacefully with each other. And even a company like CrowdFlower adopts machine learning methods, and Idibon attracts customers by combining classical techniques and the latest innovations: “ You can create your own taxonomies and use them in conjunction with the principles of machine learning and existing databases and dictionaries and”.
4. Image analysis is entering the mainstream.
Leading analyst vendors are already using image analysis technologies - for example, Pulsar and Crimson Hexagon , and the ability to analyze images using in-depth analysis was one of the key factors in acquiring IBM AlchemyAPI . A new, promising startup MetaMind , launched in 2015, states image analysis as its main feature, as it sees great prospects and opportunities in this.
5. Breakthrough in automatic speech and video analysis
“Vsekanalnaya” analytics, coupled with the complete decision-making process about buying a consumer - a favorite topic of the market. And social networks, where people most often express their opinions, are filled to the brim with video recordings. Pronounced words and non-textual elements of speech, such as intonation, speed of pronunciation, loudness and repeatability - have a certain meaning available for means of speech analysis and translation into text. We assume that in 2016, the use of such funds will significantly expand and they will begin to be used by marketers, editors and market research specialists. Also, speech analysis will most likely be used to improve the quality of computer interaction interfaces (including chat bots).
6. Advanced Emotion Analysis
Advertising specialists have long understood that when a person buys emotions, they are driven, but until recently a systematic study of reactions was out of technological reach. Depending on your perspective, start using emotion analysis or tonality analysis. The emotional state is established on the basis of an image using analysis of features and facial expressions (or from speech or text), with the goal of structuring the person's reaction to what he sees, hears or reads. Similar services for video recordings are already provided by Affectiva, Emotient, Realeyes, Beyond Verbal for speech, and Kanjoya for text; The number of users of such tools is growing rapidly and is used by many agencies, marketers, media and advertising specialists.
7. Analysis of emoji (emoticons)
We are subject to a huge number of information channels - text, images, speech, videos and likes. Why then use Emoji? Yes, because they are cool and expressive! Like # hashtags , they are more extensive forms of content. That's why Internet slang is almost dead (ROFL!) And Facebook is experimenting with reactions to emoji, and various alternatives appear, such as Line stickers . It is quite obvious that the analyst Emoji becomes absolutely necessary. The right technologies are already being used by various startups like Emogi . And although most of the projects do not go beyond the calculation and classification of Emoji, for example, Instagram engineer Thomas Dimson and research organization CLARIN.SI are engaged in such analysis, and some of them, like SwiftKey , are definitely worthy of attention.
8. Greater information retrieval from web content
With these words, we can characterize the general information trends of 2016, and it was this title that I gave an interview with Praitt Suud, a specialist in studying data in TNS . Praith notes that, “The web gives the dialogue a structure, and extracting the content gives it meaning .” Useful information is obtained by understanding the content and relationships with previously obtained information, as well as understanding the mechanism of the appearance of these relationships. So add to your toolkit a means of visualizing network content, because that's why companies like Neo4j, js, and Gephi (and that's not all) are so successful. Using a data analysis platform, such as QlikView, is an option that can be used with digital and text analytics tools: another item on the to-do list for 2016.
9. In 2016, you will consume much more automatically generated content.
The technology for automatic content generation is called Natural Language Generation (NLG) and allows you to write articles, e-mails, text messages and translations automatically, based on the analyzed text, grammar rules and context. NLG is the best solution for frequent, repetitive content, such as sports, finance, and weather reports, for example. These services are provided by Arria, Narrative Science, Automated Insights, Data2Content, and Yseop . You can also use the “services” of your favorite virtual assistant - Siri, Google Now, Cortana, or Amazon Alexa , or an automatic customer service system. Such systems fall into the Natural-Language Interaction (NLI) category, and services like Artificial Solutions will definitely be useful.
10. Machine translation will mature and settle down.
We have long been dreaming of a universal interlanguage translator (as in Star Trek films), but at least researchers in the 50s of the last century stated that the problem of machine translation would be solved within 3-5 years, but high-quality machine translation turned out to be the goal where more difficult goal than they thought. We will not argue that the full resolution of the issue is already on the horizon, but thanks to Big Data and machine analysis, 2016 (or 2017) will become that year when machine translation from the most common languages of the world will finally be good enough for most tasks. And it pleases!
Each of these trends will certainly affect us in one way or another, will directly affect - if you are analyzing text, tonality or social analytics, you consume or provide technical means - or indirectly, since the analysis of human data is already closely interwoven into the fabric of the information universe.
The general trend - more data, they are used more efficiently - to create smart “automation”, which will form the information image of the future.
Source: https://habr.com/ru/post/299886/
All Articles