AI Mapping

What is AI mapping?

When importing data to our software, corresponding database resources need to be mapped to imported rows. Most are assigned automatically thanks to multiple mapping rules. Aside AI mapping, other mapping rules require either a strict match with one of several parameters or mapping to have been done manually by a user in a previous import.

AI mapping provides resources based on imported name (and for the new Data import, country of the project). This feature simplifies the import mapping by identifying more data. It uses smart semantic recognition such as synonyms, spelling differences, and typos consideration. Strict match or previous mapping are no longer required,  and input in various languages are supported. Please note that other parameters like the construction stage or building part linked to imported material (Class) will not affect the AI mapping.

AI mapping also provides additional suggestions if you wish to replace the first match.

 

How is AI mapping improving over time?

Using AI allows to align more efficiently with your needs in the long run, by using Machine Learning. 

To be set up, our AI models have been trained with massive data, by reusing previous mapping done by our user (with correspondencies "imported name X" = "resource name X"). Preliminary work included getting rid of irrelevant data.

Our AI models then evolve and refine with each user interaction so they learn from your mapping and adapt to your preferences to provide suggestions. As users engage with the AI mapping interface (by changing an AI suggestion), the system dynamically learns from their mapping choices by being regularly trained. Then other users will benefit.*

*The country in your user settings affects the training for different regions of the world, but not the mapping you will get during a specific import.

Your feedback and suggestions are what keep making our mapping system improve all the time.

As we built a revamped import process, please find more details below on how AI mapping differs between new and old data import. 

Does AI affect other mapping rules?

To find the right data in our import mapping, the system prioritises strict match with One Click LCA ID if you are using a BIM model, EPD number or with the resource full name. It then prioritises your mapping as well as your colleagues' mapping done in a previous import. Lastly, previous mappings from users in your country and the mapping from users in other countries (global users) are also available, only on old import.

AI mapping is complementary and do not affect these rules. It substantially reduces the amount of unidentified data you could get, by not requiring strict match or previous material matching and providing corresponding resources based on name similarities. 

On old import: AI mapping happens if no mapping is found following the previous mapping from "users in the same country".

On new import: AI mapping happens if no mapping is found following previous "colleagues mapping", since "users from same country" and "global users" mapping rules are not used anymore.

AI mapping on new Data Import [BETA]

As part of our brand new data import process, we enabled AI mapping for all Building LCA tools. More info aside AI mapping on Updated Import Process in One Click LCA [BETA].

In this new import, imported name in following languages are supported by AI mapping, for any building LCA tool: English, Finnish, Swedish, Norwegian, German, French, Italian and Spanish. 
More languages might also be directly supported.

Here, AI mapping finds corresponding resources based on imported data name, but also prioritizes resources from the same country as your project country. This is not the case with old import.

You can see where AI was used in the column "Mapping basis".



You can see more AI suggestions by clicking on the cell under "Map resource", and replacing the resource assigned with a more accurate one if needed, which will help train the model.

AI mapping on old Import

AI mapping on old imports already shows an average division by 10 of the amount of unidentified data the user gets on the mapping step, for targeted tools. It only uses imported name as an input.

The feature is available for the following languages and tools:

• English: Whole life carbon assessment, GLA / RICS / Green Mark (+A2), LCA for LEED, US (TRACI), LCA for LEED, Canada (TRACI), Life Cycle Carbon - North America (metric units), Life Cycle Carbon - North America (Imperial units).
• Swedish*: LCA for Miljöbyggnad 4.0, Klimatdeklaration av byggnader - IOBC, Klimatdeklaration av byggnader.
• Norwegian*: Greenhouse gas accounting, TEK17.
• German*: LCA for DGNB (DE 2023), LCA for DGNB (DE), LCA for DGNB (Intl 2020).
• Finnish*: New YM tool (Rakennuksen vähähiilisyyden arviointimenetelmä 2021).
• French*: RE2020, Carbon energy - simplified, Carbon energy - detailed, Carbon energy - carbon alone.

To help us enhance accuracy and ensure a better mapping experience for all users, we kindly encourage you to use the “report bad mapping” feature. 

Our AI model evolves and refines with each user interaction. It learns from your mapping and adapts to your preferences to provide suggestions that align with your needs. Your feedback and suggestions are what keep making our mapping system improve all the time.