Text analytics processes unstructured text data to analyse the content of communication in social networks, webs, emails or call centres. It is used to create predictive models in combination with structured data.
The structured data analysis has been a standard part of the activities of most companies for a long time. Lately, they have seen the addition of non-structured data, namely that of the textual kind. Work with textual data is a time- and technology-intensive task. A great amount of data needs to be converted into a form that may be further processed and used in analyses. In recent days, however, technology has advanced to the degree where everything can be managed in a relatively short time.
Textual analytics is applied to the following areas:
- Classification of text documents by content;
- Analysis of communication content, eg web forum or social media posts;
- Analysis of call centre calls (combined with the “voice to text” technology);
- Identification of textual patterns and topics;
- Use of non-structured data, eg from the CRM systems, in combination with regular structured data to create predictive models;
- Fraud detection and e-mail communication analysis; and
- Identification of entities, such as company names, in non-structured data.
At Deloitte Advanced Analytics we address projects in a comprehensive manner, be they delivered separately with a focus on textual analysis or in combination with other projects as part of the analytical projects we deliver at Advanced Analytics. Our projects involve the initial collection of textual data from internal or external sources, data cleansing, transformation of data into the necessary form as well as the actual data analysis or preparation of deliverables, such as BI dashboards.
In numerous cases, textual data has been found to contribute to improved sales results or a decrease in customer churn. These days, the Czech and Slovak languages may be analysed at a high level of quality; in respect of English, this is a matter of course.
Technologies are applied to textual analytics with our clients’ needs in mind. In most cases, we use the Python, R and Java technologies or SAS and IBM tools.