When brands or companies are launching new products, users and enthusiasts take several social networks to express their feeling or sentiment regarding the newly launched items. I see most them being lost with time period due to lack of real-time analysis systems. What if we can analyze & see them with a keyword and click of a button ? I decided to make Text Analytics easy for Brands, Individual researchers & Digital Marketers based on the social media.
What it does
SnAnalyzer helps brands & marketers to analyze various social networks to get how users are reacting to their brand in a click of button. The App collects the data from various social networks (twitter, google news) based on keywords or hashtags. Then Expert AI API is being used to Analyze the text data collected and will show the overall results in a simple dashboard. Brands/Marketers/Support teams can use this to know where they are going wrong, there by making impactful decisions.
Why social networks ?
Being one of most powerful source of content, users and enthusiasts take several social networks to express their feeling or sentiment regarding the newly launched items by several brands on their product/brand pages.
Why also Google news ?
Being one of the largest news aggregators, brands will be able to find how media is portraying them with details topics being written on them, sentiment & entities mentioned.
The diagram below shows the flow of the application:
update : Added support to search with max date.
*Data Sources: *
- Google News
- Number/Count of Articles by sentiment
- Distribution of Sentiment scores
- Word Clouds By Sentiment
- Entities By Sentiment
- Similar Concepts By Sentiment
- Topics By Sentiment
- Raw Data
How I built it
I started off by installing expert.ai python client and writing code for various text analysis to get topics, entities & sentiment for given piece of text. Next, I started search for various APIs for collecting the social network data. Next, I started researching on frontend frameworks to create easy dashbaords and found it to be Streamlit. Finally, I integrated all these three to make SnAnalyzer
Challenges we ran into
- I ran into errors using expert.ai python client - 500 Response Error. But I resolved it by proper preprocessing of the text like emojis & Unicode texts. I guess expert.ai is not able to handle them at this point of time.
- Scraping data from social media is difficult on long term & as use dynamic websites which might change with time.
- This is Still existing challenge, While Power BI is working fine with python 3.5 & 3.6 scripts (having issues running python 3.7 with Power BI), the latest version of Expert.ai nl-api requires python 3.7+. So unable to complete the dashboard integration with Power BI.
Accomplishments that I am proud of
- Building an app that can use to speed up the Text Analytics for Brands, Individual researchers & Digital Marketers based on the social media. hopefully, other might find it useful too. 😃
- I only started for this hackathon 2 week ago but able to plan & built this in less than expected time, thanks to ease of use of expert.ai API
What I learned
- Using of Expert.ai API and Client
- Using Streamlit to Build Dashboards
- Scraping Data from Social Networks
What's next for SnAnalyze : Social network Intelligence for Brands
- Adding more data sources like Facebook.
- Add more filters like to search only in specified data ranges, languages & locations
- Integration of Expert.ai + Power BI
Log in or sign up for Devpost to join the conversation.