The Localization Podcast continues with episode #2. Korean Flitto to raise $33M in its IPO. Deploying multilingual chatbots worldwide. Trump’s administration hinders interpreting services. MT vs translators vs SMEs. Localization on social media.
Timestamps:
03:53 – Korea’s Flitto IPO
11:40 – Multilingual chatbots worldwide
24:00 – Trump’s administration against interpreters
27:50 – MT vs Translators vs SMEs
35:20 – Localization on social media
Andrej Zito
Hello, my dear friends of localization welcome to episode number two of the localization podcast. And yeah, now I already have a name. I didn’t have a name when I was recording episode number one. But I do have a name. Now I do have the podcast registered on several hosting podcasting services. And I also snatched the domain, the localization podcast.com. That in few years, I’m going to sell for a lot of money. Haha.
Andrej Zito
Well, today. So I’m going to share with you this one thing before we start talking about localization. And that is that I was kind of scared. I had like this interval of fear to start recording this episode. And by the way, it’s Saturday 10:49pm. And I think one of the reasons was that I wanted to do like some little preparation before I start recording, you know, so that I don’t have to read every article for the first time for this podcast, and then have like nothing to say, or just say, some bullshit that is written in the article. And that is not very relevant or valuable to you. So as I was going through articles, I kind of like, I think I got stuck on the first one, because I didn’t know like, much like, what I could add. And so that kind of brought up the fear in me that like, like, what will be the value that I will be providing In this episode, and I don’t want to be just the guy who just like read some articles, with a lot of people who know more about localization than I do.
Andrej Zito
And especially like with Slater, there are like a lot of like, kind of like high level business strategy articles. And of course, I don’t have like experience in every aspect of localization or, especially when it comes to business, then I should just, I should just keep in mind that, I mean, I just do my best and see what happens. And I just want to keep going with this. And hopefully, it will get better. I mean, I said this last time. So I don’t know, I just I just want to keep doing this. So with that said, I went over the articles from slate.com, because that’s like the only source of new publications for the localization industry. There were five new articles in the last week. And I ditched one of the articles because I think it was two. I think it was actually related to something that I could talk about. It was actually a blog post editing, machine translation.
Andrej Zito
But like the article was, like structured in a very weird way, like the style that it was written. So I decided to skip it. And I have four articles this time. Lucky for you this time, I won’t have to explain why I started localization podcast, so let’s get right into it. I should probably mention that my voice is a bit nasal these days and got sick on Monday, Tuesday. So I hope you can survive it. And without further ado, let’s get right into the first article from Slater. And that is new language industry IPO as Korea’s flinto prepares for scalability based listing. The latest language industry IPO is underway in Korea. A pen style language data startup flinto expects to raise around 33 million USD in its initial public offering when it begins trading on July 17 2019.
Andrej Zito
So that will be next week. fluido was founded in 2012 and is now listing on the development board of coal stack, the electronic trading platform of the Korea Exchange because that was launched in 1996 and is modeled on the US based NASDAQ. The Korean company conducted a demand forecast on July 1 and July 2 2019, which was oversubscribed more than 1100 times garnering interest from 1300 institutional investors According to Korean online media so what is flitter? fluido is a South Korean language data company that offers crowdsource translation, as well as professional translation services. The platform has upwards of 10 million estimated users. Much of the company’s revenue comes from selling text, voice and image data generated from translation services to data hungry tech companies such as Microsoft and Baidu.
Andrej Zito
The company also offers subtitling for YouTube content. There is a big market for selling language data and AI support services to companies that use it to power capabilities such as neural machine translation, and image or speech recognition. This market continues to grow as more companies begin to develop solutions that require good quality underlying data for machine learning applications. Latest revenues have grown rapidly over the past three years in line with this increased demand rising from zero point 36 million USD in 2015 to 4.4 million in 2018. sales for 2019 are expected to reach 6.2 million. So if they expect their IPO to value they want to raise 33 million doesn’t mean it’s going to be five times multiplied or revenue. That’s quite a lot. As photos demand forecasts as shown, interest from investors in the AI support services market is strong.
Andrej Zito
Moreover, investor confidence in language data Trailblazer apron remains sky high, with chairs in Australia listed company doubling from January to the end of June 2019. IP ns shares have shut up more than 5,000% since its IPO in 2015. And the company’s market cap currently stands north of $2 billion. Okay. So what is interesting here is that, because I actually had like no idea what this flutter company does, so I went to their website, and I saw on their homepage, they have kind of like a crowdsource translation, which I think first is machine translated. And then it goes to their pool of crowd sourced, I mean, crowd translators. An extra actually, right now I’m looking at the homepage. And it says, The crowdsourcing for the crowdsourcing service, that natural translation that fits each purpose with 10 point 3 million crowdsourcing users around the world, average response time is three minutes. And you also have the option to get a professional translator, which they call a one to one per hour translation.
Andrej Zito
So actually wondering if, because when I was reading this, and I was reading this article, the thing that kind of struck me was like if you can actually resell the data that somebody else pays for, because that’s what I understand that they’re doing, that there are people who are requesting translations. And then when these translations get finalized, I assume that then they just sell them as feed data for four to AI to train the machine translation. Is it correct, or is that not correct? I have no idea. If you have any input, please let me know. Because it’s actually surprising because like, when we had our company retreat, we were talking about something similar. And one of the questions that we raised is that whether you’re not breaching, like, contract with your client, if you would do this, not that we do it, okay.
Andrej Zito
Just let me be explicit that we don’t resell the translations that we do for our clients. Just to make it sure. So this way, actually, this struck me that this company, does it and according to the article, it seems that It’s kind of like a big business. So is it something like with Facebook that Facebook like gives you free platform, but then they resell the data about you? But in this case, like the translations that they’re reselling the data, they’re not like free like because like even the crowdsource translations are being paid for as long as I understood the the service correctly. Was there anything else important I think though one thing to note is that when I was going through these articles, it’s like it’s like machine translation is like in every other article, so I know that the first episode was quite heavy on machine translation.
Andrej Zito
So is this one so i think it’s it’s definitely something that’s that’s coming in that it’s going to change the industry, especially when it comes to translators. I think it depends on how how easy and cost effective it will be to deploy these machine translations and whether there will be just like like some like service that you can just get from like, I don’t know SDL or from from these guys or whatever you would need to start training your engine every time. second article is about chatbots multilingual chatbots the conversation has yet to get longer. article published on July 9 2019. Sales is about conversation chatbots bring this age old element of commerce to the online world. multilingual chat bots make on demand information accessible to a global to a company’s global customers creating the value in marketing as well as customer support.
Andrej Zito
Gartner forecasts that 25% of customer support and services will be integrated with virtual assistance by 2020. While no figures are available on how multilingual the bot population is, some locations are more inclined to use chatbots than others. Sorry, it’s like when I speak for a longer period of time, it’s like my voice is going away, my throat starts to get itchy. European consumers are more receptive to chatbots than those in the US. 50% of French customer hold the positive opinion on bots as opposed to only 32% of Americans. Good for the French people. Currently, chatbots are mostly used in the Leaning languages of global e commerce English fix that’s French, Italian, German and Spanish, Portuguese, Arabic, Russian, Chinese and Dutch. They have also made steady inroads into longtail languages from India and Africa to support the region’s emerging ecommerce industry.
Andrej Zito
Not all chatbots are born equal. While some are purely menu or button based and pose the least challenge. The localization wise chatbot said our keywords and context based aim to provide more fulfilling user experience Hence, the letter need to have better language skills. Menu button based chat bots are preloaded with canned responses. There are good enough in short conversations but have limited capability to respond to questions with more variables or a long tail of questions. So yeah, that’s kind of like navigating like a website technically for talking about like menu or button base chat bots because you just navigate like a kind of like tree of responses. And it’s not very difficult, you know to because basically, you just need to translate the the responses. Then the article continues.
Andrej Zito
Keyword based chatbots essentially dip into knowledge base to respond to queries from the user. After they listen to specific keywords that users type in. They make use of previously translated data from the knowledge base. However, when queries become more nuanced, Similar keywords are used, keyword based, virtual assistants can get confused. And Funny thing is that today actually before I started doing these podcasts and before I read these articles, I interacted with a first up, I thought it was my first time but ever been like in a long time. It was a chatbot from Kudo, which is our local mobile operator in Canada. And I have to say that it was it was pretty, pretty good experience. And this one was kind of like probably tailored towards younger people. Because it also I mean, on top of text responses, it also included like a GIF images of like, I don’t know, if there was like a cat based on what the chat pod response was responding or there was like, like a little penguin penguin here.
Andrej Zito
And that’s the name of the animal. There was like little penguin like running around or stuff like that. So yeah, it was a pretty good experience. And yeah, like, if you think about it, if like this, these chat bots are at the current stage, they’re kind of like, I don’t know if this article actually has some data. Oh, yeah, it actually does have these data. And this is what I wanted to say. So this will be from the article, even at this stage chatbots are expected to bring in cost savings of $11 billion in retail, banking and healthcare sectors. So yeah, so not everybody who immediately as a customer goes to customer support. Not everybody has like a very sophisticated query. And if you have like very simple queries, and you’re like lazy to find them out, and you just deploy this automated chat bot, that can dig up the information from the knowledge base and serve it to the customer, and they get their answer, like I did today, then that’s, that’s, that’s the money that you saved.
Andrej Zito
On the people who otherwise had to provide these answer manually, you would have to train them, you have to, I don’t know, rely on them how they feel and you know, like, even their responses. And I’m not sure if I’m correct here, because I have never worked in customer support. So you would have to train them, you would have to pay them. And even then, maybe their answers are different each day. While we chat bot, you have kind of like the right way to write messaging all the time. So yeah, and then once the chedworth kind of feels like lost, or they cannot provide any more support to the customer. Then you just direct the customer to the actual human agent. And that’s what even the article is written. So I’m going to finish the article chatbots that remember, try to understand meaning, and then respond appropriately need to have better language skills than their humbler peers.
Andrej Zito
People believe that anything they talk to must necessarily have human like capabilities in language, those expectations can be high. There are also the challenges brought on by people using multiple languages in the same chat or Translator lating into another language. For example, in India, it would be common to have users typing out Hindi queries in English. In Africa, people may mix English and Swahili in Kenya or French and Arabic in Algeria. plugging into a Machine Translation API can go some way by virtue of keyword detection, but may not be ready to work with life customers at scale. Even when the likes of Facebook have bots that could easily detect language given all the information in the user’s profile. It is still not enough for a fluent user experience. conversational user interfaces or C you eyes which is essentially what a chatbot is.
Andrej Zito
Need to be supported by systems powered by natural language processing and machine learning. Language designers and new job description are required to aid this learning, manage terminology and adapt a script to new locales. Designing virtual assistants to conduct a human light conversation is a tall order and requires a different set of skills than those supplied by developers. It needs to creativity of poets. And when the chatbot chat bot goes International, it is required to be similarly steered through various languages by trans creators who combine the expertise of copywriters, and translators. Additionally, domain specific fine tuning would require collaboration with subject matter experts. that technology is still relatively new and chatbots may currently be best suited to Filling in the gap before the handoff to human agents.
Andrej Zito
That’s what I was talking about before. Even at this stage chat bots are expected to bring in cost savings of 11 billion USD in the retail banking and healthcare sectors. However, with advances in artificial intelligence, human machine interactions are expected to become more commonplace, not just more efficient, and longer conversations may develop. humans and machines are currently working on that. At the end of the article, third article is about Trump, Trump’s administration and this one, this is the one that I really reduced quite a bit. Because there was like a lot of extra stuff, and I don’t think I will even be commended. But let’s see, maybe I’ll come up with something. So the title is, Trump administration’s DEA regulations regulation drive begins to impact language industry.
Andrej Zito
Man, these titles are sometimes referred me. According to a July 3, article in the San Francisco Chronicle, the US Justice Department had informed judges in immigration court about a new system that could kick kick in mid July 2019. Instead of in person, interpreters, interpreters, immigrants facing deportation in the US, we’ll get a video explaining their rights in multiple languages. Quote, after that, after that, if immigrants have questions want to say something to the judge, or if the judge wants to confirm their they understand, no interpreter would be provided the same report set. The newspaper also quoted the Justice Department official as saying that it is all part of an effort to be good stewards of the department’s limited resources, and that the new system did not represent a policy change.
Andrej Zito
So yeah, we’re saving money. The Justice Department build a move as a cost saving measure. Sources familiar with the interpreter situation say there have been ongoing issues with the budget and the contract with a primary interpreter provider, leading the administration to encourage more use of the telephone service and look for other ways to keep costs down to chronicle wrote. Second, Bart, federally regulated healthcare may also experienced similar changes. A 204 page document released on May 24 2019. by the US Department of Health and Human Services, proposes to, among other measures, replace video remote interpreting services with audio based services as it believes that additional video standards may not justify the costs.
Andrej Zito
The HHS, which is the Health and Human Services document qualifies that the proposed rule retains these access standards for persons who are deaf or hard of hearing. According to the same document, the HHS also plans to relax regulations that require healthcare agencies to provide limited English proficiency, patience, or written notice of free translation services aka taglines, which it said goodwill The US healthcare industry have an average five year burden of three point $16 billion. It’s the end of the article about Trump. I have no special comments except for saying that of course cost savings and cost savings. Okay, let’s get to the last article, please. And that will be reader posts empty and subject matter expertise feature cat outperforming lsps. And this is kind of like internal, internal internal Slater article.
Andrej Zito
And it says each week we pull them more than 10,500 subscribers to Slater’s weekly email newsletter. And by the way, as I subscribe as well, asking them to weigh in on what is top of the mind in the industry in the context of recent news. That’s interesting. At the end of June, we want to know if subject matter expertise is becoming even more important at the expense of source language proficiency. As as neural machine translation increasingly does a lot of what some called the grunt work of translating words. Many lsps are experimenting with workflows where subject matter experts to light posts editing of machine trusted content, mostly looking out for obvious errors in terminology, and apparent falls from a simpler reading of the target language without having to refer back to the source.
Andrej Zito
Our readers are fairly evenly divided over wet a better emptiness indeed drive the trade off of source language proficiency in favor of subject matter expertise with 43% disagreeing 39% agreeing and 17% saying it depends. Someone who is firmly in the yes camp is Stratos creator, Korean CEO and newly appointed board member of Polish LSP Sumo learning where you can homegirl in an interview on his recent appointment hamatos later that nnps human parity changes everything because doing words is done by a machine. However, for several use cases, revision by subject matter experts is required. So this is this is the article that I was commenting in the first episode, and I think I spent most of the time commenting on this. So what would it says right now that nmt will require revision by subject matter experts.
Andrej Zito
I i think i agree with that. What I did not agree with last time was that that translators are subject matter experts because I don’t think they are. And when it comes to source language proficiency versus subject matter, expertise. I think that subject matter expertise will matter more, the more the better. Mt. Gets. Okay, what do we have next here? But how will linguists work with all this increasingly high quality machine output? In report presented at the International annual meeting on language arrangements, documentation and publications? Oh, that’s all one event. It’s actually the acronym is i a m l a VP. That’s seven letters. It’s a forum. So this forum held in Brussels from May 27 2019. The European Union said that language technologies and ai ai, ideally take over the mundane part of language professionals work.
Andrej Zito
Now what is mundane? It’s kind of like boring. mundane My name yeah lacking interest. Yeah, yeah, I agree with that. I agree with that. The EU, one of the world’s largest buyers and users of translation, said that most of its translation services have now integrated its e translation service into their pre processing arrangements and provide an empty output to their linguists. That said, the EU still sees room for improvement in terms of sorry, in terms of usability emphasize that the future smarter cat environment is yet to be developed. Over 90% of the Slater readers we polled agree. Okay, that’s nice. So, you is also doing an empty and giving it to their linguists. But yes, there’s still room for improvement. Ah, what do they mean when they say the future smart cat environment?
Andrej Zito
How to be smarter. I have no idea. Just thinking that like, you just get the nmt output and then you just fix it. And do it in context. Somehow? I don’t know. I can’t see the future right now. I’m sorry. To finish the article, investors, meanwhile, continue to believe in the future of language services. So much so that the small group of publicly listed language service providers handily outperformed the broader stock market since the beginning of the year. leading the charge are Australia’s apron in Tokyo listed Rosetta, but rW s and key word studios to putting a solid first half of 2019. While onyekuru center and as the struggled and Zoo digital nosedived, after one of its main customers changed supply chain operations.
Andrej Zito
We asked our readers to do some crystal ball gazing and overhaul Express confidence that language industries could run is set to last for the next 12 months. And it’s the end of the article from bom, bom, bom, bom, bom. And I don’t think I have anything extra to add on top of this, ladies and gentlemen. So with that being said, I know that the last time and even in my description of the localization podcast, I said that I also cover social media. So I’m still going to be not fully delivering on that promise. But here’s the thing. So I looked at Facebook and Facebook, there’s there was like, only there’s like only one group, I think. And I think that the content there inside was the usual shit. You know? I’m not sure if it’s like, administered by people from Zillow groups, or what is the name of the company.
Andrej Zito
I know, I saw some Russian guys there. And most of the articles were, I think, for the Russian vendor. So probably not very useful. The same with Reddit. And I think I was already talking about Reddit before. So the localization Reddit is full of links to white papers and articles of some lsps, who discovered Reddit ello 2019. And by the way, I think I let me check it because I wanted to see like, what we can do with Reddit, and I made this encouraging post. When was it I did it 22 days ago. And the title of the posts was why is this subreddit dead? And still, if you go to localization Reddit, and if you sort by heart, which is I think that the fourth one, my this this post is still on the third place.
Andrej Zito
And my second question, which was how do you decide which languages translate your website into that post, which I made the second day after that is on the second place. So I’m not sure if this is just showing just for me or for other people. But yeah, localization of Reddit is pretty much net. And so with that being said, I know that I wanted to focus mainly on linked in. And I requested access to a few groups that I found on LinkedIn, but I just checked it before I started recording. For some reason, I’m still waiting for approval. So I’m not sure what that means. So I, so the only thing that I did extra was that I checked the hashtag localization on LinkedIn. There were like a lot of posts, nothing that stood up, except for one thing.
Andrej Zito
There was like this one company that had this, they had this video, which was actually produced using like, how would you call it like a video marketing? service, it’s like a, it’s like, it’s like, it’s a very nice, it’s a very nice to, I forgot the name now. And you just type in the text. And it creates like this video with all this stock footage behind, and your text will become like subtitles, and it looks nice. But it’s. But it’s not the same as if you would produce it. on your own. I think this video was like, about like introduction about their company, like what they do, you know, they bring value to the costumers, and blah, blah, blah. And they mentioned that there are like two types of translations, translation agencies, one that does like one or two languages at max.
Andrej Zito
And then there’s the second one like them, which does. I don’t know, like, they view translation from from end to end or whatever. And they started calling translation management as a service. So I did not understand it. And there was like, No comments, so I just asked, like, I don’t understand, like, what is the difference between transaction management as a service versus like, what regular lsps do because like, everybody’s here, to help the clients, you know, be a global successful brand, with successful products or services, right. So I don’t know, like, if they’re, I don’t know, like, what’s their what’s their goal. But when I checked their website, they actually like the idea that they are trying to win a first. Okay, let me start anything. When I first landing on a website, the first thing that came up was, and it surprised me because that message wasn’t part of the video.
Andrej Zito
But on their website, it says that they’re actually helping the Asian companies go international. And that’s a very nice idea, which I think I maybe even had before myself, you know, because when I was living in Southeast Asia, they wanted to focus like on like Asian customers, or maybe I just wanted to go to Philippines because I like to repeat and focus on Philippines. But But yeah, so that is kind of like a niche idea. And I like it, and they they do have some customers. So let’s see if they actually reply. And speaking about social media, I think, maybe last week, I found this, somehow it ended up on my feet on LinkedIn. I think it’s company called the word be. I think they have some, I don’t know, maybe like cloud transition management or whatever. And I was checking something.
Andrej Zito
I think it was checking like their, their their presence on LinkedIn, or was it Twitter? I think maybe it was Twitter. I don’t know right now. Yeah, I think it was Twitter. And they had like a nice post. Like, most of the time, they were not selling their stuff. But then I found like one post which was like job posting. And when I clicked the link to their website, like their their very first sentence had like a really bad grammatical error. So I told him, like, Hey, guys, like please fix this. But they did not reply to me. So that’s a very bad time. That’s a very bad time. Can only posting like a lot of content on social media. And then when you get like one comment, like, you know, like a blue moon, you just ignore it. That’s very bad. That’s very bad.
Andrej Zito
Okay, and with that being said, I’m barely breathing. And I have to drink water every few minutes because my throat gets very itchy. So I’m going to end this podcast, episode number two, right here. I know, is that valuable? I think the first one was probably much more valuable. I think that may be I should really focus on on articles or just like I don’t know, like questions like here and there on social media, where I can provide actually more content. So maybe I need to switch my focus. And again, this will probably sound like an excuse, and it probably is. I just wanted to get this done today. That is Saturday. So I can start exporting the video for YouTube during the night because it was like, one hour video and it takes like a couple of hours to export. And I want to be ready by Monday to release the podcast with all the copywriting.
Andrej Zito
So they will have to do it at the very last minute and tomorrow. I have to I have to cook and I also need to do some a little bit of work. So that’s why I wanted to do it now. And yeah, that will be it. Thank you for listening and see you. Talk to you in one week. Bye.
