Project Description

PatentsSearcher, which has started as a university project, applies cutting edge research on searching and relevance ranking algorithms.

We found that existing patents search engines just apply existing Information Retrieval algorithms to the patents database, viewing patents as regular text documents. This simplistic view leads to suboptimal search effectiveness.

In contrast, PatentsSearcher's algorithms have been created and customized for the special properties of the patents domain. This leads to the discovery of relevant patents, even if they are not obviously related (e.g., they do not have the query keywords in the abstract or title).

In addition to the high ranking quality, which is our primary focus, our system offers a rich set of features like filtering, advanced search, bulk pdf creation, Excel export and more.

This is an ongoing project and we continue calibrating, refining and improving our system on a daily basis.

We are very interested to get your feedback and in collaboration and business opportunities.

To contact us, please email to contact@PatentsSearcher.com.

Video

View the key capabilities and features of PatentsSearcher.

Frequently Asked Questions

Q: What data is indexed?
A: PatentsSearcher currently indexes all US patents and patent applications. Further, it relies on external services to query international patents and applications.

Q: Are all query words required to be in the results?
A: By default, we require all query words to appear in a section of the patent or application. That is, all query words must appear together in the abstract or the claims or the description. This behavior can be changed by unchecking "Require all query keywords to appear in the results" in the Advanced Search options.

Q: How are words separated in a query?
A: Query words are separated by spaces. If you want words to appear next to each other in a phrase, you can enclose them in quotations. E.g., the query "web search" index will return results that have web search as a phrase and also the word index.

Q: How up-to-date is your data?
A: We strive to be as up-to-date as possible. Currently, we refresh our index approximately every month. In order for you not to miss anything, check the "Possibly related patents that are too recent and not yet indexed by our system" option in Advanced Search options to display at the top of the results page a list of patents or applications that contain all the query words, ordered by date, retrieved by querying directly the USPTO for the un-indexed period (typically up to one month).

Q: How are the results ranked?
A: By default, we rank by score, which is computed by a complex formula that considers factors like frequency of words in the result, position of the words and many other secret domain-specific novel relevance factors. Related words to the original query are included in order to provide a more complete answer, unless you uncheck "Intelligently expand scope of query to locate best results" in Advanced Search options. In addition to by score, you can order the results by title, date or other fields shown under the Sort By label at the left of the results page.

Q: Can I search non-patent resources?
A: We provide links to non-patent resources, under the Search Non-Patent Sources label on the left of the results page. Currently, we provide links to Google Web Search for searching the Web and Google Scholar to search for publications related to the query. We provide two variants of these links: the Original Query simply forwards the original user query to these resources, whereas the Expanded Query adds to the query more domain-specific relevant words. Google Scholar and Google are trademarks of Google Inc.

Q: Why is the ranking sometimes changing when I resubmit the same query?
A: Our ranking algorithm leverages external sources to discover the context of a query. These external sources may constantly change, which leads to slight changes in our ranking.

Q: Is the relevance ranking the only functionality offered by your system?
A: High-quality, accurate and complete ranking of the patents and applications is the cornerstone of our system. However, we also offer an array of other innovative or common functionalities, including the following:

Publications

Vagelis Hristidis, Eduardo Ruiz, Alejandro Hernandez, Fernando Farfan, Ramakrishna Varadarajan. PatentsSearcher: A Novel Portal to Search and Explore Patents. 3rd International Workshop on Patent Information Retrieval, at ACM CIKM 2010

In the News

Intellogist, July 5 2011

People

Project Leader

Dr. Vagelis Hristidis

Contributors

Eduardo J. Ruiz

Shiwen Cheng

Abhijith Kashyap

Alejandro Hernandez

Fernando Farfan

Ramakrishna Varadarajan

Dionny Santiago

Jairo Quintana

Michael Tracy

Advisory Board

Ruben Alcoba, Esq., Alcoba & Associates, P.A., Patent Attorney

Michael J. Buchenhorner, Esq., Patent Attorney