A social media monitoring platform was build to provide insight into trends and relationships on social media and the WWW. This platform offers the user real time mentions of their brand, company or organisation. It indexed thousands of blogs, PR sites and social networks to pull back any reference to your brand and products, putting you in control of your online reputation.
The platform has a dashboard view which will alert the user to any activity surrounding your brand. This ‘buzz’ is categorised into positive and negative sentiment, enabling the user to quickly manage an appropriate response.
All data is stored in a reporting and analytical module for you to reference over time and spot trends, improvements and changes to your brand reputation
Knowledge extraction and machine learning techniques were used in order to provide useful and insightful analytics information for the user. The information presented in the dashboard allowed the user to make data driven decisions which ordinarily would have been lost in data overload.
The list of technologies involved as part of this work included:
- Java, Python, shell script
- MySQL, MongoDB
- Weka, Pandas, Scikit-Learn, NumPy, Matplotlib, Statsmodels
- JSON, XML, HDFS
- Linux/Unix server platforms
- Cloud Computing: Amazon Web Services (AWS), Rackspace OpenStack.
There were many tasks which were carried out as part of this work which included:
- Social Media Analysis
- Unsupervised learning algorithms
- Data modelling and Data mining
- machine learning
- information retrieval
- Natural Language Processing (NLP)
- Sentiment Analysis
- Gender analysis
- Large scale data processing
- Cluster computing design, configuration and management
- Big Data, SQL and NoSQL database
- API development
- batch/real-time data processing and analysis
- large structured, semi-structured and unstructured data sets
- Design and build platforms for data acquisition and processing
- Extract, clean, merge and transform data for knowledge extraction
- descriptive statistics
- Dashboard interface with integrated visualisations