Last week, my team ‘wrote’ 12 million words of real estate content for real estate professionals around the world. Each piece was optimized for search and crafted to be highly effective towards specific target markets. Oh…and no one on the team actually typed a single word.
Our Hanoi-based rapid content generation platform, Realtor Profiler, builds professional narratives for the real estate industry. While we already generate thousands of words per minute in English, our focus is broader than that. We’d like to build a multilingual platform for Asia that can instantly generate narratives in multiple languages.
Across Asia, most real estate professionals operate in their local languages. Fortunately, by combining artificial intelligence, natural language processing, and translation technology, anyone will be able to create content in multiple languages, including Mandarin (traditional), Indonesian, Korean, Japanese and Vietnamese.
In this article, I will provide a basic definition of what artificial intelligence and natural language processing are, talk about how they are changing the way we communicate, and look at the possibilities of translation technology like Google’s Translate API.
So what is artificial intelligence? Artificial intelligence, or AI, means making machines capable of performing intelligent tasks like human beings.
The term artificial intelligence has two key components:
There are different levels, or stages, of AI and although there are varying definitions I have stuck with the following three:
- Machine Learning – when an intelligent system uses algorithms to learn from experience.
- Machine Intelligence – when an intelligent system uses more more advanced algorithms to learn from experience.
- Machine Consciousness – when an intelligent system uses algorithms to learn without the need for external data like a human.
And what is Natural Language Processing (NLP)? NLP is a subset of AI; it means making machines capable of understanding and interpreting human language, both written and spoken. The objective of NLP is to make computer/machines as intelligent as human beings in understanding language. Not only human languages, but programming languages as well.
So when combined, what value do they deliver?
Today, companies generate more big data than they can humanly manage. For example, real agencies and portals retain massive amounts of data about their users’ location, performance, certifications, education, social media, and employment. NLP can analyse this information quickly to connect realtors with clients or with helpful products.
NLP is a complex field and is the intersection of artificial intelligence, computational linguistics, and computer science. It is also still an emerging technology, so a lot of possible applications are still too expensive and complex to justify the reward. Like a lot of smaller startups in this space, we are tackling a specific value proposition and rapidly evolving as the tech’s capabilities race ahead.
Realtor Profiler works with machine learning. So, first, our platform needs to understand our niche through a teachable ontology, which includes the domain model for the relevant entities and the associated expressions for the domain model. When writing a professional narrative for a residential real estate professional, our platform must, first, understand what information is most important to a prospective client.
To populate each narrative with important information, Realtor Profiler asks data-specific questions about each real estate professional and then constructs language that conveys these facts in fluent, domain-specific language.
With the same data source, Realtor Profiler designs different narratives to engage different leads on LinkedIn, portal and agency websites. Realtor Profiler customises the content based on each professional’s style, creating fluid, natural language that fits an organization’s unique voice. Also, users can tell Realtor Profiler when to pursue certain types of information through the use of story conditions.
These components, together, lets our NLP system scale to 12 million words of unique content per week without compromising quality.
Now, as I said earlier, that is all fantastic if you do business in English. But the vast majority of real estate agents in Southeast Asia work in their own languages, not in English. If we want to scale in Asia, we need to offer other languages besides English.
Enter neural machine translation technology. Google updated Google Translate with the technology in 2016, which drastically improved the quality of the platform. Google’s English to Spanish, French, and Portuguese translations will soon surpass the accuracy of a professional, human translator. The accuracy of other translations, like English to Mandarin, have improved as much as 30% since the update.
Unfortunately, Google does not process all Asian languages using the latest neural language technology. In late 2016, only Chinese, Korean, Japanese, Indonesian, and Vietnamese translations utilised the technology. The accuracy of translations is, however, constantly improving and may overtake the accuracy of human translators someday soon.
We believe that neural machine translation will change the way businesses share and analyze data all over Southeast Asia. Starting with Realtor Profiler and the real estate niche, this exciting new translation technology is completely transforming content marketing.