As part of a collaboration with leading Quebec technology influencer Bruno Guglielminetti, Semeon Analytics periodically analyzes online mentions relating to the four main provincial political parties and party leaders during the buildup to the October 1st provincial election.
Using several patented deep text analytics algorithms to analyze these mentions, various graphs are generated and help to provide our clients with a direct reading of the “pulse of the electorate” without onboarding traditional survey methods.
According to available data on these topics gathered from news sites, blogs, forums and citizen comments on public social media networks, a ‘snapshot of the pulse of the voter’ is being produced at different times during the campaign, and helps to highlight key themes that people care about, and how they feel about these topics.
To compute these relevant concept clouds and other graphs presented in this collaboration, Semeon Analytics first collects public data from millions of sources available online. By statistical inference and proximity of the researched themes (in this case the leaders and their parties), a group of several thousand documents is generated.
From the data collected, concept clouds are then produced. They provide a “qualitative” statistical representative of the frequency of relevant concepts inside the selected documents identified. The importance for each concept within the group of documents is represented by the character size. The bigger the letters are, the more important is the concept in the data set.
Additionally, the feeling associated with each concept are illustrated with a colour: green if the concept is positive, red if the concept is negative and grey if the concept is neutral or undetermined. The scores and other percentages presented in these reports are calculated from the frequency and relevance of each item within the data set.
Only excerpts are published on the twitter account of Bruno Guglielminetti (@Guglielminetti). Full reports are available upon request.
Reports’ Excerpts Published to Date
These reports are presently done only in French.
25th publication – September 30, 2018 – Analysis of feelings towards each political parties and party leaders (D-1)
24th publication – September 29, 2018 – Analysis of feelings towards each political parties and party leaders (D-2)
23th publication – September 28, 2018 – Analysis of feelings towards each political parties and party leaders (D-3)
22th publication – September 27, 2018 – Analysis of feelings towards each political parties and party leaders (D-4)
21th publication – September 26, 2018 – Analysis of feelings towards each political parties and party leaders (D-5)
20th publication – September 25, 2018 – Analysis of feelings towards each political parties and party leaders (D-6)
19th publication – September 21, 2018 – Analysis of themes towards Québec political parties after the TVA
18th publication – September 21, 2018 – Analysis of feelings towards party leaders in Quebec before and after the TVA debate
17th publication – September 20, 2018 – Analysis of feelings towards political party leaders in Québec before and after the Radio-Canada debate
16th publication – September 15, 2018 – Analysis of themes revolving around party leaders in Québec after the Radio-Canada debate
15th publication – September 15, 2018 – Analysis of feelings towards political party leaders in Québec during the Radio-Canada debate
14th publication – September 2nd, 2018 – Analysis of influencers for each political parties
13th publication – August 31st, 2018 – Analysis of themes revolving around Québec political parties
12th publication – August 24, 2018 – Analysis of influencers for each political parties
11th publication – August 21, 2018 – Analysis of feelings towards party leaders in Quebec
10th publication – August 20, 2018 – Analysis of feelings towards provincial political parties
9th publication – August 4, 2018 – Analysis of feelings towards party leaders in Quebec
8th publication – August 3, 2018 – Analysis of feelings towards provincial political parties
7th publication – July 23, 2018 – Analysis of themes revolving around Québec political parties
6th publication – July 22, 2018 – Analysis of feelings towards party leaders in Quebec
5th publication – July 20, 2018 – Analysis of feelings towards provincial political parties
4th publication – July 10, 2018 – Analysis of themes revolving around Québec political parties
3rd publication – July 8, 2018 – Analysis of domains with online mentions
2nd publication – July 7, 2018 – Analysis of feelings towards party leaders in Quebec
1st publication – July 6, 2018 – Analysis of feelings towards provincial political parties
Semeon Analytics is based in Montreal and has been developing solutions in the field of deep text analytics since 2012.
The company offers a powerful SAAS platform powered by artificial intelligence, allowing instant processing of millions of documents (public and private) to highlight statistically valid trends, feelings and intentions.
Semeon helps large and medium-sized businesses to become more customer-centric and improve their customer experience (CX) by improving their customer service, customer loyalty, organizational processes, products and services, marketing and business intelligence.
Helpful Feedback…Or Information Overload?
The voice of the customer matters. Customer input helps companies tailor the customer experience to keep their patrons satisfied and loyal. But if you run a surveying company or marketing department responsible for parsing Customer Experience (CX) data, it often feels less like useful information and more like feedback overload. Why? Because there’s so much of it.
It’s humanly impossible to scour all the data available without the help of technology. A common strategy for many companies has been to only sample data and/or to incur high costs and long delays by relying on manual analysis. But some companies have turned to machines and artificial intelligence (AI) for help. Machines process this data much faster, but all AI analysis tools aren’t created equal.
What Customers Say vs. What Customers Mean
Most AI-driven analysis relies on frequency-driven metrics. This means they try to derive meaning based on how often a word or phrase is used. These kinds of metrics completely overlook the subtleties of language to properly understand the true meaning behind customer sentiment. Tools like this ignore the complexity of language and its subtle layers of meaning. People use irony, jokes, slang, and sarcasm when they write and speak.
This makes it difficult for businesses to gather meaning from their unstructured data. For example, someone might post a comment on social media that says, “Ceramic Apple Watch looks dope as hell. I’ll be selling a kidney on eBay later.” Even a human who reads that might be confused by the meaning. So when even humans find it hard to manage language complexity, it’s understandable that machines and AI find it difficult too.
Finding Meaning Through Semantic AI
Semeon has found the answer to this difficulty. Rather than using frequency-driven, word-based metrics, Semeon instead uses a patented, AI-based semantic technique that grasps how words are used together to determine what is relevant or not. That means the Semeon analysis tools discern what is statistically relevant about what people say, helping to infer the meanings behind their words. Rather than creating word clouds, Semeon groups similar expressions together and conveys a relevancy score which it displays in a “Concept Cloud.” These clouds make it very easy to visualize customer pain points and opportunities in full sentences. They also make it easy to distinguish between positive and negative sentiments.
In the comment “Ceramic Apple Watch looks dope as hell. I’ll be selling a kidney on eBay later,” a frequency-driven tool would count the use of Dope, Hell, and Kidney and give those words equal weight. The Semeon analysis tool, however, recognizes full expressions and conveys their relevance and positive or negative meanings.
With Semeon’s visual interpretation, an analyst or executive would know that “Apple Watch looks dope as hell” indicates the customer likes the watch and this is positive. She would also know that “I’ll be selling a kidney on eBay later” indicates a negative perception that the watch is expensive. Understanding that the watch is great but is viewed as expensive is information relevant to the brand and can be acted on.
In this way, and within hours not days, the Semeon analysis tool overcomes language complexity to let your business finally hear the true voice of the customer. You don’t learn just what customers are saying; you learn what they actually mean.
Using Semantic AI to Inspire Smart Actions
Understanding customer meaning, sentiment, and intent is only the first step. The next step is knowing what action to take. Your analysis tools shouldn’t make you guess what move to make next. Instead, they should quickly deliver filtered reports with enough information that your business can make intelligent, informed decisions. Your tools need to outline actions to improve the customer experience based on what customers want.
With Semeon, your team can rapidly parse millions of data points every day to extract key insights from customer feedback. With its Concept Clouds providing relevant sentiment in plain English (much more descriptive than traditional word clouds), and its various segmented filters and detailed dashboards, analysts and executives can easily decipher and track customer sentiment and intent over time as well as for which topics they are referring to: price, staff, ease of shopping, quality, etc.
This real-time insight lets your business focus on what matters: pinpointing relevant information quickly to improve the customer experience. With Semeon, you accomplish this in hours instead of weeks or days. You’ll quickly see what’s relevant to your customers and then learn tangible actions you can take. Actionable insight will empower your organization to continuously improve the customer journey.