TABLE OF CONTENTS
- Data mining and artificial intelligence
- Understanding the fundamentals: data mining, the backbone of technology intelligence
- When data mining reaches its limits: the targeted contribution of precision AI
- TKM software: a platform built for intelligence-monitoring data mining
- From raw query to structured information: data mining in action within TKM Platform
- Combining data mining and AI makes a real difference
Data Mining and Artificial Intelligence
The challenge of technology intelligence today lies less in finding information than in sorting and organizing the data and information already available.
Patents, scientific publications, research projects, institutional reports, web sources and sector news are accumulating and spreading at a pace that far exceeds human analysis capacity.
Illustration from the cosmetics sector
In the cosmetics sector alone, nearly 2,000 new patents are published every month worldwide, reflecting constant innovation and an intense race for novelty.
Even more strikingly, artificial intelligence is seeing an information explosion: the topic is the subject of more than 50,000 scientific articles every month.
This information overload, or “infobesity,” makes it difficult to select, analyze and prioritize relevant content. R&D and Innovation teams find themselves facing an almost uninterrupted stream of information.
In this context, the challenge is no longer simply accessing information, but your ability to sort, structure and analyze it. How can you make use of these massive, heterogeneous data flows without spending your days on manual sorting? How can you avoid both information noise and blind spots?
Contrary to what the hype might suggest, generative AI has not overturned classic intelligence-monitoring approaches, and it is unlikely to do so any time soon, for at least three reasons:
- Classic methods deliver very good results in well-defined contexts with structured data. To achieve similar performance, a generative AI would need very large models or very costly retraining.
- The results of generative AI are much less interpretable than classic models, which offer genuine explainability.
- Classic models are less subject to variation and their results more reliable, unlike generative AI, whose robustness can be an issue (for example because of hallucinations).
Generative AI therefore does bring useful and interesting methodological tools. But for reasons of efficiency, as well as its high environmental cost (a subject TKM is fully committed to), this contribution comes in addition to traditional approaches, not as a replacement for them.
The solution developed by TKM is built on a rigorous approach, structured around a balanced mix combining, on one hand, the use of AI where it is genuinely useful, and on the other, data mining applied to technological literature.
It is the result of more than 20 years of recognized expertise, combined with a genuine capacity to innovate and integrate emerging technologies — not for their novelty, but for their usefulness.
By relying on structured research strategies and expert systems, supplemented where necessary by building blocks of artificial intelligence, TKM’s intelligence-monitoring and analysis software turns large volumes of raw data into information that R&D, IP and Innovation teams can actually use.
Understanding the fundamentals: data mining, the backbone of technology intelligence
Data mining, the backbone of intelligence monitoring at scale
For 25 years, data mining has established itself as the foundation of large-scale intelligence monitoring, and that’s no coincidence.
Data mining covers the full range of techniques used to explore, clean and structure large volumes of data in order to extract useful information from them. In intelligence monitoring, this means trends, thematic groupings, emerging topics or weak signals.
In our day-to-day practice of technology intelligence over more than 20 years, one observation comes up time and again: only a fraction of the information collected is of real strategic interest. Experience shows that 10 to 15% of the documents collected are actually relevant for analysis and decision-making.
The role of data mining is precisely to manage this gap between abundance and usefulness. It helps you to:
- reduce information noise without sacrificing completeness,
- eliminate duplicates,
- detect weak signals,
- structure and organize data according to relevant criteria,
- prepare a usable corpus for analysis (human and tool-assisted)
Without this upstream work, intelligence monitoring quickly becomes unmanageable, however powerful the tools used. Crawlers and other bots will soon bury you under unmanageable masses of information and turn your R&D teams away from a task that is nonetheless vital to their job: staying constantly informed of the state of the art, the competition and their freedom to operate!
→ Also read – Building a documentary archive: collecting the data useful for your technology or competitive intelligence
Expert systems: intelligence monitoring driven by the business
The information needs, the nature of the data processed and the purpose differ between an R&D team, an IP department or an Innovation and open-innovation team.
The processing useful for each of these needs is not always the same, and the workflows can be quite different.
Structuring data for an aerospace company probably has little in common with what is useful for a big pharma company or a precision-engineering SME.
Unlike magic tools and the “black-box effect” that every self-respecting analyst dislikes, expert systems are the first step in operationally deploying intelligence monitoring within the TKM platform. They are built on data-mining foundations, meaning explicit rules that can be configured by each type of user: Boolean queries, proximity operators, business filters, thesauri and classification criteria, cleaning and normalization, and so on.
This approach is particularly effective when the relevance criteria are clearly and easily identifiable: target technologies, document types, players to monitor, geographic areas, fields of application.
Expert systems then make it possible, at a lower cost1, to:
- turn a business request into a formalized search strategy,
- keep full control over monitoring scopes,
- automate the sorting, classification and distribution of information.
They offer an excellent compromise between performance, transparency and cost control, while remaining easy to adjust as needs evolve.
When data mining reaches its limits: the targeted contribution of precision AI
Not all data-processing operations can easily be handled by explicit rules alone. When defining those rules becomes complex, that is where data mining reaches its limits.
Let’s take an example!
Your business, your industry or your own strategy depends on meeting environmental criteria in the design and manufacture of your products.
It’s therefore only natural that your intelligence monitoring should prioritize, and flag to your R&D teams, any innovation, product or third-party patent (from a competitor or a startup) that meets these “environmentally friendly” criteria.
But how do you define, fully and exhaustively, all of these criteria and translate them into a rule system within your intelligence tool? That risks being complicated, if not downright impossible.
There are a thousand and one ways to describe this notion of eco-responsibility: biodegradable, chemical-free, less energy-intensive, free of rare earths, better carbon footprint, and so on.
This is where the precision, personalized AI that TKM offers you can play a decisive role, with levels of accuracy unmatched by data mining and/or a generic AI. The models are then trained on historical data annotated for the specific use case required.

At TKM, we firmly believe that in these complex cases, AI provides a genuine complementary alternative to data-mining-based expert systems. It doesn’t replace them, it enhances them.
But we are just as convinced that generic AI alone (ChatGPT chief among them) and the magic promise of a model delivering satisfying results in three clicks and four likes, simply doesn’t work!
That’s why the precision AI designed and offered by TKM relies on processing based on both generic artificial-intelligence models (classification, clustering, entity detection) and large language models (LLMs), but which also, and above all, include a prior customization step. That is when we can truly speak of precision AI (or personalized AI) applied to intelligence monitoring.
Our AI is trained on specific use cases provided by users, in order to supplement expert systems once they reach their limits.
TKM software: a platform built for intelligence-monitoring data mining
A “Data Lake” built for 360° exploration
At the heart of TKM Platform is a DataLake fed by an unmatched diversity of sources:
- global patents,
- scientific publications,
- research projects and collaborative projects,
- clinical trials,
- web sources and sector news,
- a global database of innovative start-ups and SMEs,
- a global database of academic players and major corporations.
Users can precisely target their monitoring by topic, source, author, organization or period. Specialized crawlers also make it possible to follow specific institutions or feeds, such as funding agencies, universities or research centers.
This wealth of data is essential for deploying effective intelligence-monitoring strategies and detecting weak signals at an early stage. And this, thanks to a skillful combination of expert systems and the measured use of AI.
Human support to structure the approach
The effectiveness of intelligence monitoring depends on the performance of the tools but even more so on the method. That’s why TKM supports organizations from the earliest stage: clarifying the challenges, formalizing a monitoring plan, defining research strategies.
The goal is to translate a strategic issue into an operational setup.
Depending on their needs, intelligence-monitoring teams can work independently or receive support (one-off or ongoing) to adjust and enrich the rules, scopes and processing.
→ Also read – TKM: a dual expertise bringing AI to the service of Industrial Property
Rethinking technology intelligence in the age of AI
From raw query to structured information: data mining in action within TKM Platform
Step 1: formulate a precise and exhaustive search strategy
It all starts with expressing the need.
On TKM Platform, the user is guided from a natural-language request to a structured search strategy: keyword prioritization, logical operators, filtering criteria. This step is critical to avoiding two classic pitfalls: a collection that’s too broad, generating excessive noise, or a search that’s too narrow, creating blind spots.
Step 2: clean, enrich and structure the data collected
Once the data has been collected, data mining and AI come into play.
Filtering, classification and tagging algorithms help reduce the volume of irrelevant documents and structure content according to criteria defined with the client. Where needed, trained models supplement this step wherever data mining reaches its limits.
Step 3: organize information for analysis and capitalization
Relevant information is then organized into thematic folders, populated automatically or manually. Sorting criteria are fully customizable: topic, source, date, type of player, etc.
This structuring makes analysis, targeted distribution and the long-term capitalization of knowledge easier. In the case of competitive intelligence, for example, one organizing criterion for the data could be the list of competitors, sorted by type of player or geographic area.
Combining data mining and AI makes a real difference
Getting to the point without losing the big picture
Data mining helps quickly surface high-value content within large volumes. Advanced processing also makes it possible to produce summaries, syntheses or to answer questions posed in natural language across a structured corpus.
Detecting implicit concepts and weak signals
When explicit rules are no longer enough, AI enriches data mining by detecting concepts that aren’t explicitly stated. This capability is particularly useful for spotting weak signals or emerging developments at the intersection of several fields.
Intelligence monitoring structured over the long term
Monitoring projects are updated at a frequency set by the intelligence teams. Recipients of the monitoring always have access to structured, up-to-date information, ready to use. This remains valuable over time, since the knowledge and interactions teams build up around the data corpus form a genuine intangible asset for the company.
Real-world case: using data mining for collaborative intelligence monitoring
A food-industry player wants to monitor, internationally and on a monthly basis, the scientific and technological literature (mainly patents, scientific articles and start-up news) around fermentation-related topics.
The flow is around 800 new documents to sort each month.
Despite a search strategy carefully refined with TKM’s teams, the volume of irrelevant documents remains high (our well-known 10% rule) and requires preliminary sorting before setting up automatic classification (by yeast type and field of application, in particular).
At 2 minutes per document, that represents, in theory, 3 days of work just to clean the incoming flow! And this, every single month…
That’s nearly a month and a half of work per year, for a task that is tedious and creates no strategic added value.
And that’s for just one monitoring topic, when this company needs to track at least half a dozen others…! That’s simply not workable.
In this particular case, fortunately, the criteria for distinguishing interesting information (the business rules) are fairly simple, as are the rules for classifying news within the internal knowledge and collaboration system.
For a few days of preparatory work, a system based on these rules could be deployed with a level of accuracy the manufacturer considered satisfactory.
The resulting solution, based entirely on data mining, delivers a fully operational service at minimal2 implementation cost. Processing times for sorting, cleaning, enrichment and classification operations run from a few seconds each time new information comes in. And an alert system can then, as needed, notify R&D teams of the news that matters most to them.
However, it can happen that a system based solely on business rules fails to achieve satisfactory accuracy.
That is typically the kind of situation where turning to Machine Learning and training a precision AI should be considered, in order to ultimately deliver the same service (automatic, instant processing) and usefully feed the company’s R&D or IP teams.
Conclusion
Effective technology intelligence relies above all on the smart, structured use of the right data-mining tools, driven by the business and backed by human intelligence-monitoring expertise.
Without this backbone, neither AI nor human expertise can produce reliable, usable analyses.
By combining data mining, expert systems and targeted contributions from artificial intelligence, TKM offers a pragmatic, sustainable approach to technology intelligence, aligned with the real challenges facing innovative organizations.
Do your teams need intelligence monitoring that’s more readable, better structured and genuinely usable? Contact the TKM team to discover how to put data mining and personalized AI to work for your innovation decisions.
2. Both economically and environmentally.