CSEG3060 · Research Methodology in Computer Science
Prior-art search, novelty assessment, invention disclosure, specification drafting — then filing, examination, opposition, grant and commercialisation
Patenting decisions are made during the research lifecycle, not after it. Publish first and you have already decided.
Unit V · Lecture 2 of 3 · 60 minutes · CO2 · B.Tech (CSE) Semester V
Dr. Mohsin Furkh Dar · School of Computer Science, UPES
→ / Space next · ← back · O overview · N speaker notes · F full screen
Learning outcomes · CO2
Apply the three substantive criteria of patentability, and the Section 3 exclusions that matter in computing
Conduct a structured prior-art search across patent and non-patent databases
Assess novelty claim-by-claim, and distinguish it rigorously from inventive step
Prepare an invention disclosure that gives a patent attorney everything the specification needs
Draft the structure of a complete specification, including independent and dependent claims
Describe examination, publication, opposition and grant — and evaluate commercialisation routes
Orientation
R research output · I the invention · D the disclosure · S the specification · G the granted patent. Each transition demands its own methodological discipline.
Prior-art search
What already exists?
Novelty assessment
Is any claim anticipated?
Invention disclosure
Capturing it internally
Specification drafting
The legal document
Key concepts
| Term | Definition |
|---|---|
| Prior art | Any public disclosure, anywhere in the world, before the filing or priority date, that may anticipate the claimed invention |
| Novelty | The statutory requirement that the invention must not have been disclosed in any prior art |
| Inventive step | The requirement that the claimed invention would not be obvious to a person skilled in the art |
| Invention disclosure | A confidential internal document describing the invention in enough technical detail to enable prosecution |
| Patent specification | The legally binding document filed with the office: description, claims, abstract, drawings |
| Provisional / complete specification | The provisional establishes the priority date and needs no claims; the complete follows within 12 months and carries the claims |
| Priority date | The filing date of the earliest application from which later ones claim priority |
| PSA | Person skilled in the art — the hypothetical practitioner used to judge obviousness |
| FTO | Freedom to operate — confirmation that a product does not infringe existing third-party claims |
| Patentability | Novelty, inventive step and industrial application, together |
Patentability · 1 of 2
Absolute or global novelty: not part of the prior art anywhere in the world, in any language, in any form, before the filing or priority date.
India provides a 12-month grace period for disclosures at exhibitions or by the inventor in academic publications — narrow, and not to be relied on.
A feature involving a technical advance over existing knowledge, or having economic significance, or both, that is not obvious to the PSA.
In the US, 35 U.S.C. §103, assessed against the four Graham v. John Deere factors: scope and content of prior art; differences from the claims; level of ordinary skill; and secondary considerations — commercial success, long-felt need, failure of others.
Capable of being made or used in an industry — where “industry” is read broadly under Section 2(1)(m) to include any manufacturing, service or commercial activity.
For software, this demands a specific, substantial and credible utility beyond an abstract idea.
Open-source releases that disclose the algorithmic innovation · conference publications at NeurIPS or SIGCOMM that predate the filing · public demonstrations at hackathons or product launches. Every one of these is prior art against your own application.
Patentability · 2 of 2
| Exclusion | Description | CS relevance |
|---|---|---|
| Section 3(a) | Frivolous inventions, or contrary to natural laws | Pure mathematical methods without specific application |
| Section 3(k) | A mathematical or business method, or a computer program per se, or algorithms | The governing provision for every software patent question in India |
| Section 3(m) | Mere arrangement, re-arrangement or duplication of known devices | Routine UI changes without technical effect |
| Section 3(h) | Methods of agriculture or horticulture | Computational methods may still qualify if they constitute a technical solution |
The Indian Patent Office’s revised guidelines (2017 and later) acknowledge that while computer programs per se are excluded, inventions involving software that produce a technical effect or contribute to a technical advancement are patentable.
Likely patentable — a software-implemented encryption method that improves the security architecture.
Likely not — a business method implemented entirely in software for accounting purposes.
Stage 1 · 1 of 3
Granted patents and published applications filed with national and regional offices.
Peer-reviewed journals, ACM and IEEE proceedings, theses, textbooks, technical reports, standards documents such as RFCs, product manuals.
Conference talks, posters, blog posts, open-source code commits, YouTube demonstrations, marketing material.
In many jurisdictions, including the US under 35 U.S.C. §102(a)(1), prior public use or sale is prior art regardless of whether it was ever written down.
Open-source repositories on GitHub, preprints on arXiv, and standardisation documents from the IETF, IEEE and 3GPP are now first-class sources of prior art. A commit is a publication.
Search logs must be retained: they form part of the due-diligence record required by the duty of candor before the USPTO and the disclosure obligations under Article 5 of the PCT.
Stage 1 · 2 of 3
| Database | Coverage | Strength |
|---|---|---|
| Google Patents | Global, including WIPO data | Free, full-text search, citation graph |
| USPTO PatFT / AppFT | United States | The authoritative US source |
| Espacenet (EPO) | Global, 100+ countries | Multilingual, classification search |
| WIPO PATENTSCOPE | PCT applications | The authoritative PCT source |
| IEEE Xplore | CS and EE conference and journal literature | Best for CS non-patent literature |
| ACM Digital Library | CS proceedings and journals | Best for software-related prior art |
| ScienceDirect | Multidisciplinary scientific literature | Cross-domain coverage |
| Indian Patent Database (IP India) | Indian filings and grants | Domestic filing and opposition |
Stage 1 · 3 of 3
Boolean operators, proximity operators, truncation. For a novel caching scheme:
("cache replacement" OR "eviction policy") AND ("machine learning" OR "neural network") AND ("distributed" OR "edge computing")Uses CPC or IPC codes rather than words, so it survives the vocabulary problem — a patent that calls your idea something else entirely still turns up.
Caching in distributed systems, for instance, falls under G06F12/0866 (cache hierarchies) or H04L67/568 (caching at network nodes).
Forward and backward citation chasing. Once a relevant patent is found, inspect both the references it cites and the patents that cite it, iteratively.
The closest prior art · categorisation by relevance (X, Y, A designations in EPO and Indian search opinions) · gaps the invention fills · barriers if the invention sits in a heavily patented space.
The most rigorous searches combine all three strategies with iterative refinement. A keyword-only search is the commonest reason a student concludes an idea is novel when it is not.
Stage 2 · 1 of 2
Novelty is binary. A claim is either anticipated or it is not.
Inventive step is a sliding scale, assessed against the person skilled in the art.
I is novel ⟺ ∄ p ∈ P : p ⊇ IInventive step additionally requires that I would not be reached by combining elements of P using ordinary creativity. Patentability requires both tests to pass.
If one document contains everything, you have a novelty problem and the claim is dead as drafted.
If you need to combine two or more documents to reach the claim, novelty survives and the analysis moves to inventive step — where you can argue that the combination was not obvious.
This is why the examiner’s X citations (novelty-destroying alone) are far more serious than Y citations (relevant in combination).
For software, novelty is not enough. In Alice Corp. v. CLS Bank International (2014) the US Supreme Court established a two-step test: is the claim directed to an abstract idea, and if so, is there an inventive concept that transforms it into a patent-eligible application? Pure algorithms and abstract business methods are ineligible under §101 even when they are novel.
Stage 2 · 2 of 2
Construct the broadest reasonable interpretation (BRI) of each claim
Identify the elements of each claim
Map each element to specific passages in the closest prior art
Document the element-by-element comparison
Flag any claim where a single reference anticipates all elements — a novelty failure
Flag any claim where anticipation requires combining references — this moves to inventive step
Record findings in an invention review report
Stage 3
1 Title — concise, descriptive, non-marketing
2 Inventors — names, affiliations, contribution percentages
3 Funding sources — these affect ownership and government rights
4 Problem solved — the technical problem addressed
5 Prior art known to the inventors
6 Detailed technical description — to the enablement standard
7 Best mode — the preferred embodiment at filing
8 Drawings, flowcharts, block diagrams
9 Keywords and suggested CPC/IPC classifications
10 Commercial application — industries, products, licensees
11 Public disclosures to date, with dates
12 Confidentiality statement, signed by all inventors
In academia, file the disclosure before any external public disclosure — paper submission, conference presentation, thesis deposit. Technology transfer offices review confidentially and may recommend a publication embargo.
In industry, disclosure typically goes through an internal invention management system within 30 days of conception, as employer invention-assignment policies require.
Stage 4 · 1 of 3
Establishes the priority date and describes the invention. No claims required.
Contents (Rule 15): title · a description indicating nature, purpose and components · any drawings referenced. No claims, no abstract, no detailed embodiments.
Used when the invention is at a conceptual or experimental stage, further refinement is expected, publication pressure requires an early date, or investors need an IP position established.
Filed within 12 months of the provisional. Contains the full description, abstract and claims. This is the document on which examination and grant are based.
The 12-month window is a strategic period: you hold priority for what the provisional disclosed, and can refine the claims to cover improvements made during the priority year.
The 12-month deadline is absolute. Miss it and the provisional is deemed abandoned and the priority date is lost (Section 9(1)).
Priority covers only what was disclosed. A complete specification may add matter, but cannot claim priority for matter absent from the provisional. Adding such material is new matter, and it destroys priority for those claims.
Stage 4 · 2 of 3
Concise, technically descriptive, free of trade names, aligned with classification systems.
A statement of the technical area.
Existing solutions and their limitations.
Explicit statements of the technical problems addressed.
A high-level overview of the solution.
A short explanation of each figure.
Full technical description including embodiments and best mode. Must satisfy the enablement requirement of Section 10(4) — a person skilled in the art must be able to perform the invention from it.
The legally operative section. Defines the scope of the monopoly. Must be clear, concise and fully supported by the description.
A technical summary of up to 150 words, used for prior-art retrieval and publication.
Parallel provisions: 37 CFR §1.77 in the United States, Rule 42 EPC in Europe. The structure is effectively international.
Stage 4 · 3 of 3
Independent claims stand alone and define the broadest inventive concept.
Dependent claims reference a preceding claim, add limitations, and provide fallback positions if the independent claim is invalidated.
Four claim types in CS patents:
Method — a sequence of steps, e.g. a method for training a neural network
System — an apparatus comprising components
Computer-readable medium — a non-transitory medium storing instructions that, when executed, perform a process. (Use claims are rare in CS.)
Definiteness (§112 / s.4(4)(iii)) — claims must distinctly claim the subject matter; functional language without structural support is indefinite.
Support (s.10(4)) — features absent from the description are not allowable.
Consistent terminology · no relative terms such as “large” or “fast” without quantitative anchors · an explicit definition section · disclose alternative embodiments even if unclaimed · working examples sufficient for enablement · coordinate claim scope with commercial value — too broad invites rejection, too narrow is designed around.
Anticipation by the description — the specification itself anticipates the claims
New matter — added to the complete but absent from the provisional; destroys priority
Improper dependent claims that do not further limit the parent, and lack of unity — disparate inventions in one application, objectionable under PCT Rule 13
Prosecution · 1 of 2
Unlike jurisdictions where examination is automatic, India operates an examination-on-request system. The applicant, or any interested person, must file a Request for Examination (Form 18) within 48 months of the priority or filing date, whichever is earlier.
Failure to file within this period means the application is treated as withdrawn. No extension, no remedy.
The examiner conducts substantive examination and issues an FER citing:
The applicant must respond within six months, extendable by three on a request in Form 4.
Refusals were appealed to the IPAB, now succeeded by the Commercial Courts and IP Divisions of High Courts under the Tribunals Reforms Act, 2021.
Prosecution · 2 of 2
The application is published in the Patent Journal 18 months after the priority or filing date. Earlier publication may be requested on Form 9.
Four effects: constructive notice to the public · provisional rights under Section 11A(7), allowing damages for post-publication infringement if the patent is ultimately granted and notice was given · it often triggers third-party examination requests · it opens the pre-grant opposition window.
Before publication, applications are confidential. Until then, guard against unauthorised disclosure, self-publication, and unprotected discussion at public forums.
Any person may oppose, at any time after publication and before grant, by notice on Form 7A supported by evidence.
Grounds: non-patentability under Section 3 · lack of novelty · lack of inventive step · insufficient disclosure · false suggestion or wrong mention · non-compliance with Section 8 on foreign applications.
Any interested person, within 12 months of the grant notification or its publication in the journal, whichever is later.
Grounds largely mirror pre-grant, and add non-disclosure under Section 8, fraud on the patent office, and claiming a wrong priority date.
Opposition is a defensive weapon: to block or narrow monopolies in critical technology areas, to delay or invalidate patents that threaten market position, and to test weak or overly broad claims. For the applicant, its existence is the argument for careful drafting and thorough prior-art work up front.
Grant
On successful prosecution and resolution of all objections, the Controller grants the patent. The grant is recorded in the Register of Patents and a patent number issues.
Patent number and date of grant · priority date and filing date · names of patentee and inventor · the full specification as granted · the term.
20 years from the date of filing — whether provisional or complete — subject to payment of renewal fees.
Renewal fees under Rule 80 are payable annually from the third year. Failure causes lapse; restoration is possible within 18 months with a prescribed surcharge.
Exclude third parties from making, using, offering for sale, selling or importing the patented product or process
Assign or licence the patent
Seek relief for infringement — damages, injunctions, accounts of profits
After grant
The patentee manufactures and sells. Common where the patentee has delivery capability and the invention is core to the business model.
Exclusive — sole licensee, excluding even the patentee
Sole — one licensee, but the patentee may also practise
Non-exclusive — multiple licensees
Cross-licensing — mutual, common in SEPs
Complete transfer of ownership, typically for a one-time payment, via an Assignment Deed under Section 69. Must be registered with the Controller to be effective against third parties.
Increasingly supported by academic institutions: preserves institutional ownership while enabling commercial exploitation through a dedicated venture.
Upfront fees · running royalties as a percentage of revenue · minimum royalties · milestone payments · field-of-use restrictions · geographic limitations
Cost — R&D plus procurement costs · Market — comparable transactions · Income — discounted cash flow on expected income · Real options — for uncertain but potentially high-value applications
Rapid obsolescence — software patents can be overtaken before the term expires · defensive aggregation — large portfolios built purely to deter litigation, as with the LOT Network · open-source compatibility — many are licensed on FRAND or open-source-friendly terms · standards integration — patents in H.265 or 5G NR carry significant licensing obligations.
Strategy
| Technical merit | Does it solve a real technical problem? |
| Novelty | Does it clear the prior art? |
| Commercial value | Is there a market, a licensee, a strategic motive? |
| Defensibility | Can the claims withstand invalidation? |
| Cost–benefit | Are filing, prosecution and maintenance costs justified? |
| Disclosure risk | Is this better kept as a trade secret? |
In mature IP management frameworks — for example those aligned with ISO 56005 — patent quality is preferred to patent quantity. A small portfolio of well-drafted, well-prosecuted patents outperforms a large portfolio of marginal filings.
A structured study of existing filings in a technology area, run before committing research resources. It reveals white spaces (underexplored), crowded spaces (high FTO risk), key assignees, citation clusters and filing trends.
Distinct from patentability. FTO is the ability to commercialise without infringing third-party patents, and it identifies blocking patents whose claims read on your product.
CS researchers often need FTO analysis more stringently than patentability searches, because jurisdictions that restrict patents on pure algorithms still allow patents on specific applications of them.
In academia, the Bayh-Dole Act (US) and India’s Public Funded Research Result Related IP Protection and Management Rules, 2023 let universities retain title to publicly funded inventions, subject to disclosure, election of title and march-in rights.
Consolidation
Take your group project. Run one keyword search, one CPC search and one citation chase. Write an independent claim to the template on slide 15, then try to anticipate it with what you found.
12 months provisional to complete · 48 months to request examination · 18 months to publication · 6 months to respond to the FER · 12 months post-grant opposition · 20 years term · renewal from year 3
Next — Lecture 3: The International IP System: Paris, Berne, TRIPS and WIPO; global trends; harmonisation; and filing abroad under the PCT.