Krovi: Big Data - as dream. 2nd series. Big Data negative or positive?
When are you more often angry - in the morning, afternoon, or evening? Was it possible to reduce the loss of Skolkovo for the early reduction of investments in “unsuccessful” 50-70 startups? Where to go live? You can ask other probabilistic and uncertain issues and problems - each of us has such a zillion. Often, Big Data technologies allow ... no, not to solve, but to approach the problem in general correctly, to look from the height of the Demiurge, to put the right questions and to build more optimal ways of practical solution.
In the comments to the 1st series, ServPonomarev asked an interesting question about cases of getting new meta-knowledge from Big Data. We even propose to go a little further: new technologies can allow not only to gain new knowledge, but also, in a demiurge fashion, to influence new surroundings with new knowledge.
To begin with, answer yourself one question: " How many times in social networks is more negative than positive? Is it 2, 5, 10 times? Or maybe it’s still about the same?" We all know that the social network is a real cesspool of negative, it remains to check the depth of the fall of humanity. But only grant "scientists" or in the form of "technological branch" of practical tasks can afford to engage in such nonsense.
The task of geolocating “messages” in VK, FB, Twitter, OK, Instagram and other non-FourSqure-type social networks is one of the serious commercial needs that require non-trivial solutions and massive global testing. Good leaders, of course, are trying to “get profit” from any costs. A couple of years ago, Natalya Sokolova, Brand Analytics manager, came up with the idea of ​​practical application testing a powerful new geo-module - this is how the popular regular study “Love and Hate for Regions of Russia” was born, the results of which are readily reprinted by the mass of regional and federal press and social network users governors use in their reports. The last study for the period winter 2014-2015 - br-analytics.ru/blog/lubov-i-nenavist-na-karte-rossii
')
Data for the study - more than half a BILLION of messages for 2 weeks of December from 35 million Russian authors. It is clear that the results of the study are the tip of the iceberg, and people with an analytical mind and "Mendeleev syndrome" can independently "unearth" a lot of additional interesting information comparing with other data. For example, estimate the volume of data on specific social networks from public online statistics of a social media data aggregator - pl.ilook.ru/statistics . We will definitely talk about the unique features of the social data aggregation technology in one of the series, this is a huge and interesting technological market - not for nothing did Apple “steal” $ 200 million from Twitter TopSy, Twitter, in retaliation, was forced to buy gnip, and FB marketing insights doing with DataSift .
Returning to "Love and Hate": asking the original question to hundreds of friends ("How many times more negative is more positive?") Only once heard the answer that "Positive is probably more." The study shows that positives in the statements of users of social networks in Russian social networks are MORE negative on average 4 times!
Two aspects at the end of the series:
1. A month and a half ago, the research of “scientists” of linguists on the positive attitude of humanity as a whole was published: www.pnas.org/content/112/8/2389.full - the research itself, the Russian-language interpretation of the results - lenta.ru/news/2015 / 02/10 / positive
Such studies require serious methodology and linguistic study. Brand Analytics launches the “Love and Hate” study on a semi-annual basis, because society is changing slowly and there is no need for a higher frequency for research. The speed is needed for MONITORING a changing situation and trend bursts (like “Chelyabinsk meteorite”). It takes only 1-2 days for Big Data analysts to prepare the study, because there are appropriate processing technologies for unstructured Big Data (linguistics, tonality, NER, geo-referencing), which will also be discussed in more detail later.
2. In our opinion, the most important and unexpected result : every next semi-annual research shows an increase in positivity in social networks in Russia! Positive feedback has been created: millions of people - billions of messages - Big Data technology - research - thousands of publications in social media (media, forums, social networks) - tens of thousands of comments, discussions, oaths “I will now write only positive thoughts in my own public” - growth positives of millions of people ... Man - was and is a gregarious creature :-)
Brief summary of the 2 nd series: Big Data is a data lake, but to catch a tasty fish, you need some initial information - the right questions: why did you find yourself near the lake? why do you need fish? what color, size, from what depth do you need fish? Then you get half the answer: do you need a fishing rod or spinning, sinker or float, boat or "from the shore." Enjoy your meal!
In the next series, we are sure to talk about Big Data analysts: first, our colleagues are constantly looking for those (while BD analysis is more art than skill, and hundreds who call themselves those are not enough to share themselves and the “system” or are used to working in a calmer mode (week-month-quarter, not seconds-minutes-hours); secondly, after the first series 5 requests came with questions about Big Data analytics - and we are trying to follow the signals of interest.
1st series: Big Data - like a dream
3rd series: "Obama button"
In the comments to the 1st series, ServPonomarev asked an interesting question about cases of getting new meta-knowledge from Big Data. We even propose to go a little further: new technologies can allow not only to gain new knowledge, but also, in a demiurge fashion, to influence new surroundings with new knowledge.
To begin with, answer yourself one question: " How many times in social networks is more negative than positive? Is it 2, 5, 10 times? Or maybe it’s still about the same?" We all know that the social network is a real cesspool of negative, it remains to check the depth of the fall of humanity. But only grant "scientists" or in the form of "technological branch" of practical tasks can afford to engage in such nonsense.
The task of geolocating “messages” in VK, FB, Twitter, OK, Instagram and other non-FourSqure-type social networks is one of the serious commercial needs that require non-trivial solutions and massive global testing. Good leaders, of course, are trying to “get profit” from any costs. A couple of years ago, Natalya Sokolova, Brand Analytics manager, came up with the idea of ​​practical application testing a powerful new geo-module - this is how the popular regular study “Love and Hate for Regions of Russia” was born, the results of which are readily reprinted by the mass of regional and federal press and social network users governors use in their reports. The last study for the period winter 2014-2015 - br-analytics.ru/blog/lubov-i-nenavist-na-karte-rossii
')
Data for the study - more than half a BILLION of messages for 2 weeks of December from 35 million Russian authors. It is clear that the results of the study are the tip of the iceberg, and people with an analytical mind and "Mendeleev syndrome" can independently "unearth" a lot of additional interesting information comparing with other data. For example, estimate the volume of data on specific social networks from public online statistics of a social media data aggregator - pl.ilook.ru/statistics . We will definitely talk about the unique features of the social data aggregation technology in one of the series, this is a huge and interesting technological market - not for nothing did Apple “steal” $ 200 million from Twitter TopSy, Twitter, in retaliation, was forced to buy gnip, and FB marketing insights doing with DataSift .
Returning to "Love and Hate": asking the original question to hundreds of friends ("How many times more negative is more positive?") Only once heard the answer that "Positive is probably more." The study shows that positives in the statements of users of social networks in Russian social networks are MORE negative on average 4 times!
Two aspects at the end of the series:
1. A month and a half ago, the research of “scientists” of linguists on the positive attitude of humanity as a whole was published: www.pnas.org/content/112/8/2389.full - the research itself, the Russian-language interpretation of the results - lenta.ru/news/2015 / 02/10 / positive
Such studies require serious methodology and linguistic study. Brand Analytics launches the “Love and Hate” study on a semi-annual basis, because society is changing slowly and there is no need for a higher frequency for research. The speed is needed for MONITORING a changing situation and trend bursts (like “Chelyabinsk meteorite”). It takes only 1-2 days for Big Data analysts to prepare the study, because there are appropriate processing technologies for unstructured Big Data (linguistics, tonality, NER, geo-referencing), which will also be discussed in more detail later.
2. In our opinion, the most important and unexpected result : every next semi-annual research shows an increase in positivity in social networks in Russia! Positive feedback has been created: millions of people - billions of messages - Big Data technology - research - thousands of publications in social media (media, forums, social networks) - tens of thousands of comments, discussions, oaths “I will now write only positive thoughts in my own public” - growth positives of millions of people ... Man - was and is a gregarious creature :-)
Brief summary of the 2 nd series: Big Data is a data lake, but to catch a tasty fish, you need some initial information - the right questions: why did you find yourself near the lake? why do you need fish? what color, size, from what depth do you need fish? Then you get half the answer: do you need a fishing rod or spinning, sinker or float, boat or "from the shore." Enjoy your meal!
In the next series, we are sure to talk about Big Data analysts: first, our colleagues are constantly looking for those (while BD analysis is more art than skill, and hundreds who call themselves those are not enough to share themselves and the “system” or are used to working in a calmer mode (week-month-quarter, not seconds-minutes-hours); secondly, after the first series 5 requests came with questions about Big Data analytics - and we are trying to follow the signals of interest.
1st series: Big Data - like a dream
3rd series: "Obama button"
Source: https://habr.com/ru/post/287062/
All Articles