Demystifying innovation: harnessing natural language processing to comprehend organizational innovation

The Schumpeterian theory of recombinant innovation posits that innovation is the process of combining existing ideas to synthesize novel ones. The three studies of this dissertation aim to enhance our understanding of organizational innovation by using the Schumpeterian theory of recombinant innovation as the foundation principle and leveraging natural language processing. The first study examines the relationship between the gender diversity of inventors within firms and innovation outcomes, specifically recombinant innovation. Recombinant innovation involves combining two or more disparate fields of knowledge to create new ideas. However, the measurement of firm-level recombinant innovation remains challenging. This study proposes an information artifact called SPaRK (Semantic Patent Recombinant Knowledge) using word embeddings and clustering to measure recombinant innovation using patents. The study finds that gender diversity among a firm's inventors has a positive impact on recombinant innovation, overall innovation, female innovation, and financial outcomes for the firm. The second study proposes a framework to assess the combined innovation potential of firms and validate it by evaluating its impact on the post-M&A outcomes of such firms. We leverage the "Value Innovation Theory" for the framework design, which suggests that innovation should be value-driven rather than technology-driven. Specifically, this study proposes value embeddings (i.e., value-based vector representations) to encode the value of firm patents in the semantic-value space. An information artifact called CoMET (Combined Maximum Embedding Total) is constructed from value embeddings of patents to measure the combined innovation potential of firm pairs. We evaluate the proposed information artifact in the context of mergers and acquisitions (M\&A). This study finds that M&As with higher combined innovation potential of constituent firms, as measured through CoMET before mergers, have higher innovation quantity, quality, and knowledge-sharing post-merger. The third study focuses on identifying combinations of existing ideas that could lead to potential innovations (i.e., synergistic combinations) that are valuable for firms. The study proposes a Synergistic Combinator (S-combinator) framework and considers the identification of new synergistic combinations as a prediction task. By leveraging the design science paradigm and an experimental setup, five kernel theories and meta-designs are evaluated to identify the ideal S-Combinator framework design. This framework is suitable for guiding R\&D initiatives within an organization and external patent procurement.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Business Administration

Grantor

University of Illinois Urbana-Champaign

Year dc:date

2025

Author dc:creator

• Ramaraju, Naveenkumar

Contributors dc:contributor

• Pant, Gautam
• Subramanyam, Ramanath
• Ghoshal, Abhijeet
• Pant, Shagun

• Innovation
• Recombinant Innovation
• Natural Language Processing
• Text Analytics
• Value Embeddings

Statement dc:rights

• © 2025 Naveenkumar Ramaraju

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/130015

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F130015