From its inception, Media Tenor has used professionally trained analysts for its media content analysis. Even though there has been a great deal of investment in the area of artificial human intelligence, no software has yet been developed that can compete with the human brain. Data generated by software analysis may save costs, but the accuracy of such data is not high enough to predict people's behavior or to research reputation risks.
Human Content Analysis vs. Software based Content Analysis
1. International Media Analysis: Using native analysts that understand historical, political, and cultural backgrounds produces highly reliable and qualitative data. |
1. International Media Analysis: When software analysis is conducted across multiple languages and cultures, accuracy problems occur, since the software exists in a limited number of languages and automated translations are unreliable. Software can not evaluate cultural aspect at all. |
2. Human analysis is consistent, as it discloses how results were obtained. |
2. Software analysis results in Black box measurement, since software does not reveal the details of the analysis. |
3. No problem with special types of expression. |
3. Texts using irony, symbolic, metaphors, colloquial language, satire, or sarcasm can not be correctly analysed at all by software, which is incapable of understanding them. |
4. When new topics, sources, mediums, etc. appear, analysts assign new codes in the system and inform researchers. |
4. Software analysis is unlikely to identify text or words not programmed into the software dictionary. These are simply omitted. |
5. Has no problems providing both relational and conceptual analysis. |
5. Provides only conceptual analysis—keyword presence and keyword counts. Some of the most advanced software can provide certain limited relational analysis . |
6. Every single information is encoded and considered. Only through this approach is Agenda Setting Research possible. |
6. Information about clients and topics of analysis are taken into consideration, but the rest of the information that influences the public is not considered. |
7. Analyses are time consuming but highly qualitative. Speed is achieved by employing more analysts. |
7. Although software is not capable of substituting for human judgment, it is capable of assisting researchers in mastering a large unknown corpus very quickly. |
8. Flexible and able to change quickly, as people are trained daily on new issues. |
8. Less flexible, requiring reprogramming in order to implement new issues. |
9. Since all texts are analysed according to a codebook reflecting daily reality, media Impact Analysis and Awareness Threshold identification is possible. |
9. Software does not take all information into account, and thus it is not possible using software to identify AwarenessThresholds or to analyse the impacts of certain issues on your reputation. |
10. Text is analysed by professionally trained analysts. |
10. Automated content analysis makes very arbitrary associations between words and phrases. |
11. Text identification is 100 percent as all articles are read and properly understood. |
11. Automated approaches to information filtering, which rely on the occurrence of a given set of keywords to identify possibly relevant texts, are limited by the fact that they involve no actual understanding of the input texts. |
| Consulting based on qualitative analyst coding and research. | Consulting based on software output. |
For this page following literature was used:
M. R. Lissack, A New Twist for Content Analysis
K. Neuendorf, The content analysis guidebook
W. Lawrence Neuman, Social Research Methods
C. Newbold, O.Boyt-Barett & Van Den Bulck, The media book
K. Krippendorff, Content Analysis – an introduction to its methodology
W. Evans, Computer-Supported Content Analysis: Trends, Tools, and Techniques
© 2008 Media Tenor International