Technology Commercialization Activation Model Using Imagification of Variables

Round 1

Reviewer 1 Report

Congratulations on the article. I find it a very interesting subject because usually, Technology Transfer is treated as an intangible. Still, this article indicates that this process can be measured and controlled.

1 - What is the ideal database size to train and obtain a reliable model? For example, can I train the CNN in a short period, such as 2019-2020?

2 - Has the model been tested with data obtained in the United States? Can this model be replicated in other countries based on this data? It would be interesting to answer this question in the discussion section. If it is possible to gather data from several countries, one would have a global model to level the global technology transfer processes.

Author Response

Dear. Reviewer 1

We wish you only good things. We can further advance our research with your professional review. The revised thesis was written after careful consideration and consultation by all authors. The following is the response content according to the review.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper deals with the commercialization of technology based on the transfer of patents. The authors define transfer as the transfer of technology from R&D entities, such as universities, research institutes, and companies to others. Forms of technology transfer include licensing, mergers and acquisitions (M&A), and the transfer of intellectual property rights. The paper focuses on the transfer if patents. The aim of the paper is to develop a model for the prediction of technology commercialisation.

The authors use a dataset of patents on artificial intelligence at the USPTO, all in all 15,193 patents between 1989 and 2016, thereof 4,137 transferred patents.

For their model, they use numerical data, first of all usual data on patent quality whereof the majority is weak except forward citation and family size. In addition, they use text data and look at typical pattern.

All in all, the suggestions for the development of a prediction model are convincing, but various questions remain:

- How are transferred patents identified? The exact definition is important for the validity of the model.

- For the quantitative data, forward citations are used. How were the forward citations determined?

- How are litigation data identified?

- What are the elements in text data which refer to transfer or commercialisation? In general, text data in patents only refer to technology. Does it mean that also the text data refer to technology?

- What is the target of the prediction? Identify technologies which are likely to be transferred?

If all data used refer to technology and value, does this mean that the prediction is finally based on technologies that where frequently transferred in former times, e.g. if AI patents referring to language were frequently transferred in the last years, patents referring to language will be frequently transferred in the next years.

To summarize, the authors seem to be exclusively oriented on model techniques, but less on the content of patents and the problems of transfer and commercialisation. Therefore, it is not clear who will use their model for what purpose.

Various clarifications are needed.

Author Response

Dear. Reviewer 2

We consider ourselves very fortunate to have received a review from you. Your professional views and comments have given us the opportunity to take our research to the next level. All the authors have discussed the contents you mentioned, and the paper has been appropriately revised. Please check the attached file.

Author Response File: Author Response.docx