Before I begin, I want to demonstrate the power of big data. I have 6 things on the slide that I know that gamers like. Tell me how much you like these things by clapping.
Why do we clap? We clap because something is cool. We clap because we feel that we have to defend our beliefs, our choices, our souls. We clap because something is totally awesome an d unexpected. We clap because we want to give the performer a kindness in their moment of vulnerability. And you can measure applause. Just be judging the noise in the room, I can tell that this group would be more interested in certain types of games than others. And you’d probably like it better if it were offered with beer rather than Lindsay Lohan.
In the modern world, a like, a share, a retweet, a reblog is a clap. Whenever we friend or follow, plus 1 something or link-in, we’re applauding. And in the same way that I was able to determine that you all like Zelda better than Farmville by listening to your claps, we can do the same with all the smack talking that’s going on online. And this can lead to some valuable insight.
What kind of valuable insights? Here are a few things online conversations can tell you. But online conversations can tell you even more than just a survey. I could have asked you all to take a survey and calculated the results. But the problem with that is that you’d only answer the questions that I have and it wouldn’t lead me to even question the things that I don’t know or that you don’t know you know. Like how you talk to one another and the slang that could be incorporated into our translations.
There are a lot of conversations happening online and we aggregate these conversations using our social listening tool called SM2. The conversations that are happening are sometimes treated. In other words, certain conversations have more influence than others and are therefore weighted for relevancy. When we pull these conversations, we are looking for very specific key words as they relate to certain processes within a customer journey. For the purposes of this presentation, I’m focusing on one type of conversation which indicates the liklihood to buy a product. These conversations look like this.
When the conversations are gathered, and we’re talking about millions of outputs in some cases, we are able to use an algorithm to determine sentiment which is plotted out over a period of time. In this case, we plotted the campaigns for the Cityville 2 release by Zynga. Some key campaign highlights give us an indication of how effective these milestones were to driving revenue. What’s interesting about this is that each country told a different story.
If were to illustrate how effective these campaigns were by clapping, Germany and France would sound like this. Maybe even a golf clap. While Italy, would be off the charts like this. They LOVE Cityville2. Meanwhile, Spain…well, they would sound like this. Not impressed at all.It’s good to know that people like or dislike your product in a country. And most big-data firms can give you these charts. The challenge is understanding WHY Spain is not as impressed with Enrique Iglesias’ abs as Italy. And in order to do that, you need people who can actually understand what the data is saying, in the language it was written in.
You can aggregate all the data you want, but without a linguistic resource or machine translation to provide context, data is just data. The really important information that can help you make better business decisions, better translations, is lost in translation simply because there is so much to translate, where do you even begin? Thankfully, machine translation can quickly expedite the process for discerning which information is relevant and which information is not. Linguists can than translate professionally any artifact that supports the general conclusions we’ve uncovered from the data.
Smack Talk, Big Data, and Localization
Smack Talk, Big Data,and LocalizationLiesl LearyMarch 2013
Online conversations = unstructured data (a lot of it)
Meaningful insights=*which markets should we localize into?*what kind of slang do players use?*what do gamers want?*what will gamers pay for?*what kind of gamers do I have?*how can we get more registrations?
Examples of online conversationsPage 6EmotionalRationalBehaviouralCustomer Relevance Scores: I think….“Blizzard is too complicated in they way they talk to me”“Cityville doesn’t use my language”“I need to get more rewards by adding more friends to thisgame”“I should share my prize with other friends”“I’m not trying hard enough”“It’s hard to understand the rules of this game”“I am wasting time by playing this game.”Brand Commitment Scores: I feel….“Envious for not having achieved more in this game”“Bored waiting to progress in this game”“Stressed, isolated and frustrated”“At the mercy of luck & have no control over my score”“Inadequately trained to play”“That the sequel is not living up to promises”“That more friends would make this more enjoyable ”Product Commitment Score: I do….“Spend my money to purchase this game”“Spend my time doing mundane repetitive tasks rather thandoing what I’m paid to do”“Everything myself”“Ignore a lot of what I’m sent”“…not believe a lot of what X says”“…not make the most of the tools available to me”“…not invest time in my development”
From Data to Meaning9Big Data•QUANTIFIES sentiment through the CustomerCommitment Framework•Generates more than 50K words per project/perlanguageMT•Typical time constraint is 2.5 weeksfor translation for any volume ofwords. MT achieves this goalLinguists• QUALIFIES thedata formeaning• Professionallytranslates whatis important
Your Presenter10Liesl Leary, SDL Global Solutionslleary@sdl.comTwitter: lieslramaLinkedin: www.linkedin.com/in/lieslleary