Mining Text for Meaning
Sources report that more than 80% of the world’s data is in text form. Many companies, in order to grow, need to find meaning in this textual information, and research into text analysis has been ongoing for several decades.
To mine meaning from text, it is first structured: Each sentence is grammatically analyzed for parts of speech and syntactic relationships. Different analyses have different goals. The following list is not inclusive:
- Clustering text
- Categorizing text
- Extracting concepts
- Analyzing the sentiment (i.e., determining the tone of the writing)
- Summarizing
- Deciphering relationships among elements within the text
Existing text mining systems allow for methods that are statistic and probabilistic in nature. While these approaches have proven useful, they have inherent limits that do not provide the sophisticated understanding of the text that a human analyst provides.
Current research into text analysis aims to move beyond statistical and probabilistic relationships to provide an understanding of how humans process, generate, and comprehend language, thereby allowing for more symbolic approaches, approaches that “understand” relationships between ideas described in paragraphs.
In the near future, more complex search queries—queries based on ideas and themes rather than key words alone—will yield more focused information retrieval. Other benefits of advanced text analysis approaches are applicable to:
- Business intelligence
- Automated translation
- Information sharing
- Criminal investigation
At AGS Analytics, we are close to rolling out a text analysis system. We’d like to speak with you about your company’s needs now so that we can tailor our software to your goals. Please call or e-mail us. Initial consultation is free—our intrinsic energy is a bonus.