Solution to Help Companies Patent their Inventions, License Technologies, and Address Litigation Challenge
2 likes1,307 views
The document discusses AI Patents, a technology search engine designed to identify related prior art from vast patent databases, addressing challenges posed by varying textual descriptions of inventions. It outlines the competitive advantages of AI Patents' algorithms that outperform traditional search engines by providing more relevant results and examples of its effectiveness across various applications, including patent litigation. The company aims to improve technology search processes using its advanced semantic analysis capabilities, ultimately benefiting users in efficiently locating pertinent patents.
Solution to Help Companies Patent their Inventions, License Technologies, and Address Litigation Challenge
- 1. AI Patents – Patent, Product, Technology Search A Revolutionary Engine for Technology Search Liat Belinson, CEO Liat.belinson@aipatents.com ICIC 2015, 19 - 20 OCTOBER . NICE
- 2. AI Patents Motivation How to identify related prior-art from millions of patent documents? Large textual domain: An invention can be described in many different ways Where to search? Prior-art cited by the patent examiner is usually from a different technology classification than the application In about 60% of patent application rejections, the application and prior-art cited are from different technology classifications (IPC) Key words: Patent documents typically include technical phrases that differ textually, but may be related scientifically E.g., PMG (permanent magnet generator) and BLDC (brushless DC controller) are related scientifically, but differ textually Incentives: Patent prosecuting attorneys have an incentive to differ their clients’ inventions semantically from the most related prior-art to mitigate the risk of rejection
- 3. Patent Search Query: “the driver of the vehicle is notified by the presence of pedestrians, animals or any other environmental or weather related hazard affecting the roadway in real time” Competitor A Search query uses long complex syntax Not sorted by relevancy Too many results AI Patents Competitive Advantage
- 4. Free text query Relevant results Sorted by relevance AI Patents Competitive Advantage Patent Search Query: “the driver of the vehicle is notified by the presence of pedestrians, animals or any other environmental or weather related hazard affecting the roadway in real time”
- 5. Search Query: a golf ball finder system which helps golfers identify the position of their lost golf balls; for example, if the golfer hits the ball out of bounds, the user is alerted to the location of the ball by means of an LCD display or sound output (search across issued US Patents) AI Patents vs. Competitors’ Test
- 6. 1. US 7,207,902 - Method and apparatus for locating and recording the position of a golf ball during a golf game 2. US 5,910,057 - Golf ball with distance and locating system 3. US 6,113,504 - Golf ball locator 4. US 6,634,959 - Golf ball locator 5. US 6,524,199 - System for locating a golf ball Search Query: a golf ball finder system which helps golfers identify the position of their lost golf balls; for example, if the golfer hits the ball out of bounds, the user is alerted to the location of the ball by means of an LCD display or sound output (search across issued US Patents) AI Patents vs. Competitors’ Test
- 7. 1. US 7,207,902 - Method and apparatus for locating and recording the position of a golf ball during a golf game 2. US 5,910,057 - Golf ball with distance and locating system 3. US 6,113,504 - Golf ball locator 4. US 6,634,959 - Golf ball locator 5. US 6,524,199 - System for locating a golf ball Search Query: a golf ball finder system which helps golfers identify the position of their lost golf balls; for example, if the golfer hits the ball out of bounds, the user is alerted to the location of the ball by means of an LCD display or sound output (search across issued US Patents) 1. US 5,013,046 - Multiple mark golf ball and playing method 2. US 5,163,686 - Practice mat for golfers 3. US 6,309,130 - Wood-type golf heads provided with vertical grooves on hitting surface 4. US 4,413,827 - Scaled-down golf course game 5. US 6,015,353 - Golf swing stroke-error-signaling sensor 1. US 7,641,565 -Method and apparatus for detecting the placement of a golf ball for a launch monitor (note: this patent is for flight path monitoring) 2. US 4,334,707 - Golf ball retriever (note: this patent is for scooping a golf ball, e.g. out of a body of water) 3. US 7,497,780 - Integrated golf ball launch monitor 4. US 7,207,446 - Detachable golf ball display 5. US 5,589,628 - Golf ball striking device 1. US 7,515,980 - Sound outputting apparatus and sound outputting method 2. US 8,340,323 - Sound outputting apparatus, sound outputting method, sound outputting system, and sound output processing program 3. US 5,988,314 - Sound output system 4, US 5,144,670 - Sound output system 5. US 8,204,241 - Sound outputting apparatus, sound outputting method, sound output processing program and sound outputting system Provided the same search query, AI Patents produces higher quality references compared to its semantic search engine counterparts AI Patents vs. Competitors’ Test Vendor 2 Vendor 1 Vendor 3
- 8. Patent Search Query: “ a bicycle is a personal transportation device with wheels and pedals that transfer from pedalling through a chain to the rear wheel and has handlebars for steering” Key Words: not bike, not motorcycle, not bicycle, not cycle ” AI Patents Competitive Advantage
- 9. Our Solution AI Patents draws from a novel and comprehensive analysis of related scientific and technology concepts (e.g. from learning from USPTO and EPO patent examiner reports), to measure the similarity of different patent documents. The system can identify technologies by their conceptual relatedness in a way that is more robust than traditional keyword/Boolean- based representations. Rigorous testing validates that our algorithms substantially outperform existing search engines. AI Patents is poised to become a critical part of the technology search process as it dynamically – and automatically - becomes more educated over time AI Patents’ semantic search engine learns from the accumulated and collective expertise of the USPTO and EPO Examiners. who make 102/103 rejections everyday.
- 10. Our Solution – Examples of AI Patents’ Dictionary Search Input DDR* Search Input encode* Example 1 Example 2
- 11. Implemented Technology Multiple Application Spaces – Generic Semantic Technology for different areas In the next slides, we demonstrate how this technology excels in various application spaces as: Patents search PubMed search (largest biotech and medical articles archive) Products descriptions search Company descriptions search Resumes search search by full job descriptions with the best candidates at the top
- 12. We examined several patent suits and loaded a free description of the technology into AI Patents’ search engine with the following results: Alcatel-Lucent vs. Microsoft – MP3 ($1.5B), 2003, patent filed at 1997. Our system identifies the relevant patent in location #5 (without any filtering on years). NTP vs. RIM ($612.5M), 2001. 5 patents are involved in the litigation case. Our system identifies those patents in locations #1, #2, #5, #9, #20. Intellectual Ventures vs. Chris Crawford, 1993. Our system identifies the relevant patent in location #2. Microsoft vs. Barnes & Noble, 2011. Our system identifies the relevant patent in location #2. Alcatel-Lucent vs. Microsoft – Microsoft Outlook (500$M), patent filed in 1986. Our system identifies the relevant patent in location #14 (see more detail and results in the next two slides). Sample Cases If the infringing companies would have performed Freedom-to-Operate searches using AI Patents’ engine, they may have avoided lawsuits and spending millions of dollars in the process
- 13. In order to test on an existing case, we uploaded a description taken from the "Technology" slide in the following PowerPoint deck www.cs.berkeley.edu/~tlavian/.../Alcatel-LucentvMicrosoft.ppt The following text was inputted into AI Patents’ search engine: “Invention Box: Quantizing noise – approximation of continuous range by values by relatively small set of discrete values, Invented method and apparatus to produce quantized audio signal using interpolated scale factor” Boolean Search: audio*, signal, noise AI Patents immediately found the relevant patent ranked 14th in the results. How much would that have saved Microsoft? sample case: Alcatel-Lucent vs. Microsoft
- 14. sample case: Alcatel-Lucent vs. Microsoft
- 15. Query: “Iron is an essential mineral for the function of all body cells and is absorbed at the apical surface of enterocytes to be transported by ferroportin, the only known iron exporter, across the basolateral surface of the enterocyte into circulation. Inflammation from IBD interferes with iron absorption by causing an increase in hepcidin, a peptide hormone synthesized in the liver that inhibits ferroportin activity. A deficiency of iron can have a significant impact on a patient's quality of life” PubMed Boolean queries are extremely complex Queries complexity results in EMPTY output AI Patents Competitive Advantage (PubMed)
- 16. Query: “Iron is an essential mineral for the function of all body cells and is absorbed at the apical surface of enterocytes to be transported by ferroportin, the only known iron exporter, across the basolateral surface of the enterocyte into circulation. Inflammation from IBD interferes with iron absorption by causing an increase in hepcidin, a peptide hormone synthesized in the liver that inhibits ferroportin activity. A deficiency of iron can have a significant impact on a patient's quality of life” Cited Article placed into 1st position Highly Relevant Articles Found Sorted by relevance AI Patents Competitive Advantage (PubMed)
- 17. Search Query: cloud based platform that allows businesses to better manage their CRM processes from lead generation to request for proposals, from expense tracking to employees' sales performance records Search Engine Results Retrieved 1. Filenet Business Process Manager – IBM 2. Cognos Customer Perfomance Sales Analytics – IBM 3. Expense Management JD Edwards Enterpriseone – Oracle 4. Tivoli Availability Process Managers – IBM 5. Commerce on Cloud – IBM About 341,000 results (1.10 seconds) 1. Best Lead Management Software | 2014 Reviews of the ... 2. Top CRM Software | 2014 Reviews of the Best Systems… 3. CRM & Cloud Computing To Grow Your Business ... 4. CRM Software | Free Pricing & Feature Comparison 5. 50 Best Sales Management Tools/Software – Docurated 6… AI Patents Competitive Advantage (Products Search)
- 18. X solutions increase revenue for online merchants through automated, real-time customer interaction at the point of customer conversion. Based on behavioral targeting of site visitors, Smart Offers from X increases conversion rates and captures more revenue from your existing site traffic. Purpose: Applying AI Patents’ intelligent search engine to company profiles of 100 companies Search Query: 1. Runa - Runa allows online merchants to more than double the conversion rate of visitors. A one percentage point increase in conversions from visitors to sales among all the existing online merchants is equivalent to them spending approximately an additional $7 billion on online advertising at current efficiencies. 2. Sweet Tooth - Sweet Tooth increases sales by converting first time visitors into loyal customers that crave your brand. These loyal customers come back to your store, purchase frequently, and promote your brand to their friends & family. Online businesses can create loyal brand promoters by automatically engaging their customers using Sweet Tooth’s advanced loyalty platform. Over 3,000 online businesses are using Sweet Tooth to automate the communication to, and rewarding of, their customers. 3. Zephyr Financial Technologies - Web Based platform creates a completely new 360degree environment to access and analyze global financial data. The system design, architecture and marketing solution fuses together real time financial data (pricing, charts, analytic tools etc.) via specific data streams, aggregated content and Business Intelligence (BI/FI) Applications into a dynamic and customizable interactive interface. 4. ShopLogic - ShopLogic gives e-commerce merchants an intelligent way to optimize and manage promotions on their site. ShopLogic lets merchants promote intelligently and determine what to offer, who should get them, and how much to discount. The best promotion is delivered for each shopper, resulting in more margin and sales for The following search results – all in the top 10- were the most relevant to the search query AI Patents Competitive Advantage (Company Profile)
- 19. Testimonials “It is certainly a lot easier to search with your tool than freepatentsonline.com or the USPTO” “We were able to identify companies making comparable products. This led us to several potential manufacturing Partners” An Actual Case Study “AI Patents helped us understand the competitive landscape better than any search we'd done before. It was fast and efficient and the results were incredibly relevant to the work that we were doing. I would highly recommend their product and will use them again” One of AI Patents’ first customers saved more than $300K for not buying a “wrong” patent: - originally used a competitor application did not find relevant patents. -Then, turned to AI Patents, and on his first search found a pre- dated patent with the exact technology as the one he considered acquiring. Now our satisfied paying customer Case Studies & Testimonials
- 20. Liat Belinson E-mail: Liat.belinson@aipatents.com Phone: (919) 599-5512 Contact Information
- 21. Our Solution – Source of information In most cases before the USPTO, a patent application will receive a prior art rejection under 102 or 103. In this case (application # 12/942,047), the Examiner cited US 6,747,268 as an anticipatory reference. Therefore, there must exist some link, or correlation, between the words in the 12/942,047 app and the ‘268 patent.
- 22. 6900000_102 5443941 6900000_102 5733712 6900000_102 5991081 6900000_102 6747268 6900002_103 6103305 6900002_103 5767018 6900002_103 6103305 6900002_103 6500756 Our Solution – Process of building the dictionary Extracting references from 102 and 103 rejections to create the list below: Grouping references into worlds Produce semantic analysis Generate technical dictionary