Digital Innovation as Pockets of Discursive Cohesion

Summary

This lecture explores the ongoing research project on 'Digital Innovation as Pockets of Discursive Cohesion,' examining how employees at Volvo Cars describe their skills and experiences in a free text format. Utilizing NLP techniques, the project analyzes the unstructured data to identify patterns and clusters in language, revealing insights into digital innovation practices and their diffusion across the organization.

Takeaways

• 🌐 The lecture discusses a research project on 'Digital Innovation as Pockets of Discursive Cohesion', focusing on how digital innovation is discussed and represented in language within an organization.
• 📝 The research is based on a dataset from a SAP plugin used by Volvo Cars employees to describe their skills, experiences, and tasks in free text format, allowing for natural language expression without predefined categories.
• 🔍 The dataset is analyzed using Natural Language Processing (NLP) techniques to understand the language used by employees, which can reveal underlying rationales and institutionalized expressions.
• 📚 The script introduces basic concepts of NLP, such as vector representation of texts, term frequency, and the importance of mathematically representing text for computational analysis.
• 👥 The analysis combines individual employee texts into 'documents' representing organizational units, allowing for the comparison of language use across different units within the company.
• 📊 A heat map is used to visualize the cosine similarity between different organizational units, showing the linguistic closeness or distance based on the used vocabulary.
• 🔑 The research identifies 'pockets' of discursive cohesion, where language use becomes aligned, potentially indicating areas of digital innovation within the organization.
• 🔍 The project explores the hypothesis that digital innovation initiatives can be captured in language, as people invent and adapt language when working together on innovative projects.
• 📈 The script describes a method for creating a more precise lexicon by identifying unique terms within the organization that are not part of everyday language, using a set operation with a reference lexicon from an NLP library.
• 🔬 The use of Term Frequency-Inverse Document Frequency (TF-IDF) helps to highlight terms that are significant within specific documents but not as common across all documents, aiding in the identification of meaningful language patterns.
• 🚀 The findings suggest that digital innovation can be observed in how language is used and evolved within organizational clusters, indicating the potential for new practices and the reinvention of existing ones.

Q & A

What is the main topic of the virtual lecture?

The main topic of the virtual lecture is 'Digital Innovation as Pockets of Discursive Cohesion,' which explores the relationship between digital innovation, discourse, and language in the context of an organization.

What is the significance of the data set used in the research project?

The data set is significant because it comes from a plugin to SAP, specifically from employee profiles, where employees describe their skills, experiences, and tasks in free text format. This unstructured data allows for a natural expression of ideas and is valuable for analyzing language patterns related to digital innovation.

How does the research project utilize Natural Language Processing (NLP)?

The project uses NLP to process and analyze the large volume of unstructured text data. It involves vectorizing the text, aggregating text into documents for organizational units, comparing texts using cosine similarity, and applying clustering to identify patterns and similarities in language use.

What is the purpose of vector representation in NLP as discussed in the lecture?

Vector representation in NLP is used to mathematically represent text, allowing for the comparison of texts based on the frequency and presence of specific words or tokens. This simplifies the analysis of large text data sets and helps in identifying similarities and differences between documents.

How does the concept of 'discursive cohesion' relate to digital innovation within an organization?

Discursive cohesion refers to the alignment of language use when people work together and innovate. In the context of digital innovation, it suggests that when employees with different backgrounds and roles come together and use language in a cohesive manner, it can lead to the development of new practices and the reimagining of technology, which is a core aspect of digital innovation.

What is the role of 'statements' in the context of the research on digital innovation?

Statements, as defined by the researcher, are words or expressions that carry particular meaning within a specific context, such as digital innovation initiatives. These statements can be unique to the initiative and serve as a fingerprint of the practice, helping to identify and understand the nature of digital innovation within the organization.

How did the researchers create a more precise lexicon for their vectorization process?

The researchers created a precise lexicon by using a set operation to identify words or tokens that are important within the organization but unknown outside of it. They compared the vocabulary from a large NLP library, which served as a proxy for everyday language, with the specific language used in the employee profiles, resulting in a focused lexicon of statements.

What is the significance of the heat map in the research?

The heat map is a visual representation of the cosine similarity between different organizational units based on their language use. It helps to identify clusters of units that use language in a similar way, indicating potential areas of digital innovation and areas where practices may have been reinvented or generalized.

How does the research project address the challenge of analyzing a large data set?

The project addresses this challenge by employing computational methods of NLP, which allow for the efficient processing and comparison of large volumes of text data. Techniques such as vectorization, aggregation, and clustering are used to identify patterns and make sense of the data at scale.

What insights can be gained from the clustering of organizational units based on language use?

Clustering based on language use can reveal groups of organizational units that are working on similar topics or challenges, indicating potential areas of digital innovation. It can also highlight the diffusion of technology and practices across different parts of the organization, showing how digital innovation is being adopted and adapted.

What is the potential application of the findings from this research project?

The findings can be used to better understand how digital innovation emerges within organizations, identify areas where new practices are being developed, and inform strategies for fostering innovation. It can also provide insights into the diffusion of technology and the reinvention of traditional practices in the context of digital transformation.