Artificial Intelligence (AI) In Telecoms & Media
Artificial Intelligence (AI) In Telecoms & Media

Artificial Intelligence (AI) In Telecoms & Media

Machine Learning and Deep Learning is now being used in the telecommunications and media industries. But why group media companies with telecommunications? The trend of big telecom companies buying media companies for their content is continuing and even accelerating. (Fast Company, 2017)

Machine Learning and Deep Learning are a growing and diverse fields of Artificial Intelligence (AI) which studies algorithms that are capable of automatically learning from data and making predictions based on data. Machine Learning and Deep Learning are two of the most exciting technological areas of AI today. Each week there are new advancements, new technologies, new applications, and new opportunities. It’s inspiring, but also overwhelming. That’s why we created this guide to help you keep pace with all these exciting developments.

Whether you’re currently employed in the mining industry, or just pursuing an interest in the subject, or working with us at Produvia, there will always be something here to inspire you!

AI Research in Telecoms

In order to take advantage of exponential power of artificial intelligence, research is the first place to look. Luckily, we have done the hard work and compiled our favourite research papers as it relates to telecommunications industry.

Detecting Fraudulent Calls

Telecommunications has experienced dramatic expansion due to the rise of mobile phone adoption. However, with increasing number of mobile phone users comes increasing amount of phone fraud. Mobile communication fraud is common since it is easy to get a subscription using fake ID and mobile terminals are not bound to physical locations. These factors allow fraudsters to profit with relatively low risk of getting caught. Mobile phone fraud is defined as the unauthorised use, tampering or manipulation of a mobile phone or service.

  • Detect fraudulent calls in mobile phones by analyzing the user’s calling behaviour using machine learning. (S. Subudhi, 2016)

Predicting Customer Churn

Churn models aim to identify early churn signals and recognize customers with an increased likelihood to leave voluntarily. “Over the last decade there has been increasing interest (in fraud detection) studies in areas including telecommunication industry.” (T. Vafeiadis, 2015)

Many machine learning algorithms are used to tackle the churning prediction problem. These methods include: Artificial Neural Networks, Decision Trees learning, Regression Analysis, Logistic Regression, Support Vector Machines, Naive Bayes, Sequential Pattern Mining and Market Basket Analysis, Linear Discriminant Analysis, and Rough Set Approach.

It’s now possible to classify customer’s segmentation based on Customer’s Age Group, VIP Status, Spend Status and Customer Length of Service using machine learning and data mining. (C. Dullaghan, 2017)

By predicting when customers will churn, telecommunications and media companies can save money.

The telecom industry suffered losses of around £953 million [$1.56 billion USD] in 2011, a based on an average loss of 2.4 per cent against the total operator reported revenue of £39.7 billion [$64.9 billion USD] (Union Street, 2014). In 2016, Telus lost 1.21% of subscribers per month due to churn. [23]

Predicting Customer Experience

Real-time understanding of customer experience and satisfaction is one of the key competitive advantages of telecommunication companies. Telcos deal with large amounts of information generated by it’s users every day. For example: “mobile data traffic is forecasted to reach 24.3 Exabytes per month by 2019, which corresponds to a 10-fold growth from 2014 to 2019”. Such large and fast mobile data makes it harder for telcos to extract customer insights in a timely enough manner to react to potential causes of poor customer experience. [24]

It’s now possible to classify customer experiences based on data feeds, customer care calls, spatial distribution, and temporal distribution using a supervised learning approach of Restricted Random Forest. [24]

Pratical AI In Telecoms

There are many companies that are already using AI, machine learning and deep learning in their products and services. Here are some of our industry favourites.

Telus

  • Telus uses machine learning to monitor “noise rise from more than 20,000 cell towers to increase service and device availability and mean time to repair (MTTR)” using Splunk . TELUS provides wireless services across Canada with a network of over 20,000 cell towers. Early identification and remediation of difficult-to-detect network incidents is crucial to maintaining customer satisfaction. TELUS uses Splunk machine learning to “reduce cell tower downtime and increase service availability.” [6]

CenturyLink

  • CenturyLink offers Data & Analytics Services as part of it’s IT Services & Consulting [18]

Comcast

  • Comcast holds PHLAI — “a technical conference for engineers and professionals interested in and working with machine learning and artificial intelligence.” [7]
  • Comcast used H2O's [17] machine learning to improve X1 by XFINITY [19] features, improve customer care, improve customer experience, create more resilient & reliable products [8], prevent avoidable truck rolls (an appointment at a customer’s home or business), predict trending content to offer personalized suggestions and browsing options for TV audiences, develop customer experience metric, improve product resiliency [9]
  • Comcast Applied Artificial Intelligence Research team uses NLP, computer vision, and machine learning (with a focus on deep learning) to invent “the technological foundations for the Xfinity experiences of the future” [10]
  • Comcast uses Apache Spark [20] to “detect anomalies in customer activity that may indicate service interruptions” by analyzing “usage clickstreams and contract events such as telesales and emails” [11]

AT&T

  • AT&T Big Data Research uses machine learning, data visualizations to “invent the new technologies and build platforms which empower AT&T’s Big Data initiatives” [12]
  • AT&T Intelligent Services and Platform Research uses artificial intelligence, human/machine interaction to “conduct research that empowers the next generation of autonomous and interactive services” [12]
  • AT&T Labs uses artificial intelligence, networking and data analytics to develop “machine learning and massive data modeling platforms” [13]

Verizon

  • Exponent by Verizon [21] offers “software solutions built specifically to meet the needs of the global carrier community”. Exponent offers Artificial Intelligence Platform “designed to assist carriers to unlock and monetize their wealth of data through the application of advanced machine learning techniques, deep analytics, and artificial intelligence” [14]
  • Verizon uses Apache Hadoop [22] and Apache Spark [20] for big data infrastructure [15]

Nokia

  • Nokia offers machine learning-powered customer experience solutions via Customer Care Solutions [16]. Solutions include: reduction of time a customer is on a phone, improvement of solving cases on first contact, escalations reduction, call reduction due to successful self-help/IVR, reduction of propensity to call (PTC), repeat call reduction, improvement of NIPS/recommend to others, truck rolls reduction due to self-care, and improvement of call prediction [16].

AI Ideas for Telecoms

Want to build something new? There are way to apply AI to the telecommunications industry. Here are some ideas worth exploring.

  • predict customer churn (T. Vafeiadis, 2015)
  • predict customer experience [24]
  • predict downtime, including problem and location [25, 26]
  • predict content popularity [27]
  • predict service interruptions/disruptions [28, 11]
  • prevent avoidable truck rolls (an appointment at a customer’s home or business) [9]
  • predict trending content to offer personalized suggestions and browsing options for TV audiences [9]
  • develop customer experience metrics [9, 8]
  • improve product resiliency [9, 8]
  • improve customer care [8]
  • recommend order of videos shown for genre rows by combining personalization with popularity, and identifying and incorporating viewing trends over different time windows ranging from a day to a year (Personalized Video Ranker) [29]
  • recommend top pick videos based on the top of the catalog, rather than the entire catalog (Top-N Video Ranker) [29]
  • recommend trending now videos using short-term trends, such as an interest in holiday movies or films driven by weather events (Trending Now) [29]
  • recommend movies to resume based on intent to resume watching, rewatch, or abandon something not as interesting as anticipated (Continue Watching) [29]
  • recommend next movie to watch based on previously watched movie (Video-Video Similarity) [29]
  • recommend rows of videos based on mood, different time, occasion, or member of household (Page Generation: Row Selection and Ranking) [29]
  • recommend movies based on evidence items (ex. deciding whether to show that a certain movie won an Oscar or show the member that the movie is similar to another video recently watched by that member) (Evidence) [29]
  • recommend what to watch based on search keywords for videos, actors, or genres (Search) [29]
  • recommend Live TV programs to watch
  • recommend On Demand programs to watch
  • recommend Highlights to watch
  • recommend programs to record

Want us to help you with an AI project? Chat with us at produvia.com

References

  1. Dullaghan, C., & Rozaki, E. (2017). Integration of Machine Learning Techniques to Evaluate Dynamic Customer Segmentation Analysis for Mobile Customers. arXiv preprint arXiv:1702.02215.
  2. Vafeiadis, T., Diamantaras, K. I., Sarigiannidis, G., & Chatzisavvas, K. C. (2015). A comparison of machine learning techniques for customer churn prediction. Simulation Modelling Practice and Theory, 55, 1–9.
  3. Subudhi, S., & Panigrahi, S. (2016). Use of fuzzy clustering and support vector machine for detecting fraud in mobile telecommunication networks. International Journal of Security and Networks, 11(1–2), 3–11.
  4. Telecoms Fraud | Minimizing The Risk | Union Street. (2014). Union Street. Retrieved 11 July 2017, from http://www.unionstreet.uk.com/telecoms-fraud-minimizing-the-risk/
  5. Kepes, B. (2017). Splunk ups the machine-learning ante. Computerworld. Retrieved 11 July 2017, from http://www.computerworld.com/article/3125517/big-data/splunk-ups-the-machine-learning-ante.html
  6. Machine Learning | Splunk Enterprise | Splunk. (2017). Splunk. Retrieved 11 July 2017, from https://www.splunk.com/en_us/products/splunk-enterprise/features/machine-learning.html
  7. Presentations, C., & Conduct, C. (2017). Comcast Labs. Comcast Labs. Retrieved 11 July 2017, from https://phlai.comcast.com/
  8. Ambati. (2015). H2O World — Machine Learning at Comcast — Andrew Leamon & Chushi Ren. Slideshare.net. Retrieved 11 July 2017, from https://www.slideshare.net/0xdata/h2o-world-machine-learning-at-comcast-andrew-leamon-chushi-ren
  9. Operationalizing Machine Learning at Comcast. H2Oai. Available at: http://www.h2o.ai/wp-content/uploads/2017/03/Case-Studies_Comcast.pdf. Accessed July 11, 2017.
  10. Research — Comcast Labs, DC. (2017). Comcast Labs, DC. Retrieved 11 July 2017, from http://dclabs.comcast.com/research/
  11. Dinsmore, T. W. (2016). Disruptive Analytics. Apress.
  12. AT&T Labs Research — Our Research. (2017). Research.att.com. Retrieved 11 July 2017, from http://www.research.att.com/evergreen/what_we_do/research.html
  13. What’s Next at AT&T Labs? AI Set to Revolutionize the Network | AT&T. (2016). About.att.com. Retrieved 11 July 2017, from http://about.att.com/innovationblog/next_att_labs
  14. Verizon Launches Exponent, a New Technology and Business Venture Designed to Accelerate Growth for Global Carriers. (2017). Exponentplatforms.com. Retrieved 11 July 2017, from http://www.exponentplatforms.com/verizon-launches-exponent
  15. Large-Scale Machine Learning at Verizon. MMDS. 2014. Available at: http://mmds-data.org/presentations/2014/srivastava_mmds14.pdf. Accessed July 11, 2017.
  16. Customer Care Solutions. (2016). Nokia Networks. Retrieved 11 July 2017, from https://networks.nokia.com/solutions/customer-care
  17. H2O.ai. (2017). H2o.ai. Retrieved 11 July 2017, from https://www.h2o.ai/
  18. Data & Analytics Services | CenturyLink. (2017). Centurylink.com. Retrieved 11 July 2017, from http://www.centurylink.com/business/enterprise/it-consulting/data-analytics.html
  19. X1 by XFINITY® | Comcast Cloud DVR. (2017). X1 by XFINITY® | Comcast Cloud DVR. Retrieved 11 July 2017, from https://www.xfinity.com/x1
  20. Apache Spark™ — Lightning-Fast Cluster Computing . (2017). Spark.apache.org. Retrieved 11 July 2017, from https://spark.apache.org/
  21. Exponent Platforms, a Verizon Company. (2017). Exponentplatforms.com. Retrieved 11 July 2017, from http://www.exponentplatforms.com/
  22. Welcome to Apache™ Hadoop®!. (2017). Hadoop.apache.org. Retrieved 11 July 2017, from https://hadoop.apache.org/
  23. TELUS Annual Report 2016 — Welcome. (2017). About.telus.com. Retrieved 12 July 2017, from http://about.telus.com/investors/annualreport2016/?lang=en
  24. Diaz-Aviles, E., Pinelli, F., Lynch, K., Nabi, Z., Gkoufas, Y., & Bouillet, E. et al. (2015). Towards Real-time Customer Experience Prediction for Telecommunication Operators. Arxiv.org. Retrieved 12 July 2017, from https://arxiv.org/abs/1508.02884
  25. Telus down? Realtime status and problems overview . (2017). canadianoutages.com. Retrieved 18 July 2017, from http://canadianoutages.com/status/telus
  26. BV, S. (2017). Downdetector on the App Store. App Store. Retrieved 18 July 2017, from https://itunes.apple.com/ca/app/downdetector/id816223770?mt=8
  27. Content Popularity for Open Connect — Netflix TechBlog — Medium. (2017). Medium. Retrieved 18 July 2017, from https://medium.com/netflix-techblog/content-popularity-for-open-connect-b86d56f613b
  28. British Columbia Wildfires. (2017). Forum.telus.com. Retrieved 18 July 2017, from https://forum.telus.com/t5/Service-Status/British-Columbia-Wildfires/ta-p/74358
  29. Gomez-Uribe, Carlos A., and Neil Hunt. “The netflix recommender system: Algorithms, business value, and innovation.” ACM Transactions on Management Information Systems (TMIS) 6.4 (2016): 13.
  30. Why 2017 Will Be A Huge Year For Telecom And Media Mergers. (2017). Fast Company. Retrieved 24 August 2017, from https://www.fastcompany.com/3068696/why-2017-will-be-a-huge-year-for-telecom-and-media-mergers

If you know of any machine learning applications related to telecoms or media companies, please leave a comment below.

If you have any questions about artificial intelligence or the future of telecoms or media, feel free to message us at produvia.com


Artificial Intelligence (AI) In Telecoms & Media was originally published in Produvia Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.