![APPENDIX
APPENDIX 1: PATENT COUNT BY COMPANY
[Write a paragraph on how we came up with each of the competitors]
APPENDIX 2: 3M’S TOP 20 PATENTS BY CLASS COMPARED TO FOUR COMPETITORS
[write a paragraph to explain the tables above]
Company Count % of Total
GE 12,528 38%
Siemens 7,329 22%
Honeywell 6,186 19%
3M 3,485 11%
Dow 2,581 8%
Du Pont 373 1%
Johnson & Johnson 296 1%
Bostik 56 0%
Total 32,834 100%](https://image.slidesharecdn.com/faaf8b1f-9249-4339-815c-72d187949237-161229210039/85/3M-Technical-Summary-Part-2-2-320.jpg)
3M Technical Summary - Part 2
- 1. TECHNICAL SUMMARY Big Data for Competitive Advantage / DSBA 6100 EXECUTIVE SUMMARY Patent Data Overview We analyze 3M’s US patent portfolio relative to seven competitors with topic modeling, k-means clustering and network analysis. The dataset includes about 33,000 patents for the eight companies (Appendix 1). Relative to the selected competitors, 3M has a competitive advantage in areas like stock materials, synthetic resins, optical systems and adhesives (Appendix 2). Using cosine similarity on the class distribution by company, 3M’s patent portfolio is most similar to Bostik, Dow and Du Pont, who are predominately synthetic materials and chemistry companies. General Electric and Siemens focus more on energy and data & processing patents; however, these companies compete with 3M in areas like surgery and stock materials patents (Appendix 3). Topic Modeling & K-Means Clustering Next, we use Topic Modeling (Latent Dirichlet Allocation or LDA) to identify five common topics in the patents’ abstracts. We label the five topics as: Synthetic Materials, Chemistry, Energy, Electrical and Data & Processing. (Appendix 4). We then use K-Means clustering on the topic probabilities to create five patent clusters, each corresponding to a LDA topic. Most of 3M’s patents (75%) are in Synthetic Materials and Chemistry patents (see left). Network Analysis Third, we examine shared Patent Title “bigrams” to analyze the relationship between patents using network analysis. Cross-polination is a major tenant of 3M’s current innovation strategy. Synthetic Materials and Chemistry patents have the strongest “cross-pollination” as they share many bigrams in classes like compositions, adhesives, abrasives and synthetic resins (see graph on the left). On the other hand, 3M’s Electrical and Data & Processing patents are largely disconnected from the major component. 3M has a competitive disadvantage in these patents as GE, Honeywell and Siemens dominant the market. 3M may need to either exit or expand (organically or acquisition) these patent groups in the future. Moving Towards a new Innovation Strategy Using our results and our competitive industry analysis (see our Industry Analysis memo), we are working towards recommendations for 3M’s innovation strategy. We intend to frame our recommendations using the topic clusters as well as provide enterprise-level recommendations (e.g. management or organizational changes). 1. 3M is unique. 3M has a range of competitors per patent type but no competitor across all of 3M’s patent types. 2. 3M’s competitive advantage is in Synthetic Materials & Chemistry patents. 3. These patents exhibit high cross-pollination (shared topics), a leading tenant of 3M’s current innovation strategy. Patent Counts by Topic Cluster & Company 3M’s Patent Portfolio thru Shared Bigram
- 2. APPENDIX APPENDIX 1: PATENT COUNT BY COMPANY [Write a paragraph on how we came up with each of the competitors] APPENDIX 2: 3M’S TOP 20 PATENTS BY CLASS COMPARED TO FOUR COMPETITORS [write a paragraph to explain the tables above] Company Count % of Total GE 12,528 38% Siemens 7,329 22% Honeywell 6,186 19% 3M 3,485 11% Dow 2,581 8% Du Pont 373 1% Johnson & Johnson 296 1% Bostik 56 0% Total 32,834 100%
- 3. APPENDIX 3: 3M COSINE SIMILARITY BY CLASS DISTRIBUTION Cosine similarity is calculated by taking the vector of the class distribution for each company and calculating a cosine similarity for any two companies. For example, assume there are five patent classes and three companies: 3M, Dow and General Electric. Assume each company’s patent portfolio is distributed by class like: We then take each patent portfolio distribution vector (e.g. for 3M: 40%, 30%, 20%, 10%, 0%) and plug each into cosine similarity to get a sample cosine similarity matrix: Cos Sim 3M Dow GE 3M 1.00 0.93 0.57 Dow 1.00 0.48 GE 1.00 % Patents 3M Dow GE Stock Material 40% 60% 20% Abrasives 30% 30% 10% Surgery 20% 5% 20% Optics 10% 5% 10% Turbines 0% 0% 40% Total 100% 100% 100%
- 4. APPENDIX 4: FIVE TOPICS IDENTIFIED AND LABELED FROM LDA Labeled Topic Name Topic Word Cloud (top 15 words by probability) “Synthetic Materials” “Chemistry” “Energy” “Electrical” “Data & Processing”