“Semantics, syntax, and context.” You need all three to turn text into possible business decisions, according to Seth Redmore, CMO at Lexalytics, Inc.
Lexalytics has been surfacing the intentions and sentiments in the global social conversation since 2003, and brands using major social relationship management platforms or PR services like Sprinklr, Trendkite, Falcon Social, or Cision may be leaning heavily on Lexalytics’ text mining capabilities without even realizing it.
Indeed, the customer base for the Massachussetts-based text analytics vendor is quite specialized–as we’ll see. But it certainly contends to add major value to solutions brands and agencies are considering as part of their marketing tech stack.
On Premise and SaaS Solutions
Venerable by marketing tech standards, Lexalytics originally came packaged as an on premise, multi-lingual textual analysis engine. Known as Salience, and currently in its sixth iteration, Lexalytic’s signature product was the first commercial sentiment analysis engine, and it’s still an important part of the company’s proffer, processing an estimated three billion documents per day.
Salience integrates with in house listening apps to mine everything from tweets to call center records and survey responses for intention and sentiment. It uses Wikipedia categories as a preconfigured taxonomy; automatically identifies and extracts named entities (individuals, companies, etc); summarizes relevant longer texts; and scores texts for sentiment (not overlooking emoticons, emojis, and hashtags).
Aspects of the solution are customizable too. Salience can be taught to recognize and extract lists of specialized terms and expressions, and can return entites under your chosen, normalized name–for example, it will return references to the President, to Obama, and to Barack as instances of “Barack Obama,” if that’s your preference.
Lexalytics came to the cloud arguably late in the day, acquiring the cloud-based text and sentiment analysis engine Semantria in July 2014. The purchase brought Semantria full circle. It had been developed within Lexalytics, and powered by Salience, but had become an independent entity in 2011. As a SaaS-based solution, it can be deployed almost instantly via API or an Excel plug-in, and scaled to meet demand. Like Salience, it’s customizable for industry specific search terms. It’s billed monthly.
What a Feeling
Sentiment analysis takes social listening a step beyond knowing that people are talking about a brand or product. It measures the attitudes expressed in conversation, and scores them for strength. As Lexalytics defines it, sentiment analysis “is the process of detecting the contextual polarity of text. In other words, it determines whether a piece of writing is positive, negative or neutral…A common use case for this technology is to discover how people feel about a particular topic. For example, do people on Twitter think that Chinese food in San Francisco is good or bad?Analyzing tweets for sentiment will answer this question for you. You can also learn why people think the food is good or bad, by extracting the exact word indicating why people did or didn’t like the food. Example: ‘too salty.'”
Do people on Twitter think that Chinese food in San Francisco is good or bad?Analyzing tweets for sentiment will answer this question for you.
It takes only a moment’s thought to realize that the deep problem facing sentiment analysis is context. Everyone–almost everyone, I should say–has learned to detect things like humor, sarcasm, irony and hyperbole in face-to-face conversations. It’s harder to do so, of course, in written exchanges–emails, for example. When it comes to reviewing countless tweets by complete strangers, it ought to be almost impossible. “President Obama’s doing a great job. Right.”
The key, as Redmore explained to me, is contextualization. You have a shot of detecting sarcasm or irony in the tweets or updates of a complete stranger if you know what they’ve said about the topic in question in the past. Indeed, if you think about it, a statement made in isolation can’t really be sarcastic: it requires the contrast with sincerity. You get a good feel for what’s meant by an update saying “I love the Apple store,” if it follows an earlier update saying, “I stood in line at the Apple store for three hours yesterday.”
Improving contextualization, first for groups or clusters of online voices, then for individuals, remains an ongoing challenge for Lexalytics.
Another opportunity the company is looking at is ad targeting, said Redmore. Sentiment can be a basis for audience segmentation and for honing personas.
Look Elsewhere for Video Analytics
Lexalytics may hold pole position when it comes to analysing text-based content for emotion and opinion, but digital marketers are increasingly looking to video- and image-based campaigns to engage audience on sites like Instagram, Pinterest, YouTube, and–above all–Facebook. Mining the vast universe of unstructured data for sentiment would seem to be a mountain Lexalytics would welcome the chance to climb.
We’re convinced that text is still changing and growing
Not so. “We’ve bounced this around internally, and made the strategic decision to stay entirely focused on text,” said Redmore. “We have partners who can extract metadata from spoken words or video, and we’ll work with them, but the challenge is dealing with text and getting that right.” After all, it’s not like there isn’t enough text to go around. “We’re convinced that text is still changing and growing so quickly that someone has to remain focused on it. We want to be the company to come to for text.”
Lexalytics is also not in the data business. It deploys across data aggregated by other solutions, but it stores no data once the analysis is over. “Even in our cloud service,” said Redmore, “we throw data away.”
Specialized Audiences
As for who should come to Lexalytics, brands and agencies considering the purchase of broader services–social media management, social listening, digital marketing and PR–might do well to learn whether they come with Lexalytics or alternative text and sentiment-mining strategies baked in.
As for Lexalytics direct selling, Redmore explained that customers fall into two primary classes. The first is the very large tech companies “that have a strong belief in their own data and want total control over what they do with it”–including the analytics. Microsoft and Ciso are examples. The second is smaller, boutique market research firms and specific departments within larger companies, with a regular need to augment data like survey responses or search results with sentiment analysis.
A free trial of Lexalytics’ SaaS solution is available here.
|
Company Name: Lexalytics Headquarters: Boston, Massachussetts Categories: Analytics, Social Media Marketing/Social Listening, Customer Experience Management/Voice of Customer Describe Yourself: Processing billions of unstructured documents every day globally, Lexalytics is the industry leader in translating text into profitable decisions. Lexalytics deploys state-of –the-art cloud and on-prem text and sentiment analysis technologies that transform customers’ thoughts and conversations into actionable insights. The on-premise Salience and SaaS Semantria platforms are implemented in a variety of industries for social media marketing/social listening, reputation management and voice of the customer programs. Salience is Lexalytics’ on-premise text analytics engine that processes over three billion documents per day. Semantria is the world’s largest cloud-based text mining service which analyzes over eight billion documents per month (or over 100 billion a year) through its API service. Main Competitors: IBM/Alchemy API, Clarabridge, Bitext, Metamind Website: http://www.lexalytics.com/
|
