|
|
| (11 intermediate revisions by the same user not shown) |
| Line 1: |
Line 1: |
| | The following table lists the current AI features in Compass. <br/> |
| | <br/> |
|
| |
|
| <pre style="color: red" >
| | {| class="wikitable" style="margin:auto" |
| These functions are currently still in the pilot phase.
| | |- |
| Individual functions, interfaces and scope may change in the future.
| | ! Feature !! Description !! Supports at !! AI Type !! License |
| </pre>
| | |- |
| | | | [[EN:Text_search#AI_-_Full-Text_Search|AI Full-Text Search]] || AI-supported semantic search || Search || Internal || Light/Advanced |
| The AI features in the [[EN:DetailView AI|detail view]] are available by default for all users with an Advanced license. <br/>
| | |- |
| | | | [[EN:Text_search#AI_-_Add_Synonyms|AI in Full-Text Search]] || Automatically find synonyms || Search || External || Light/Advanced |
| ---- | | |- |
| | | | [[EN:Highlighting#AI_-_Synonyms|AI in Highlighting]] || Automatically find synonyms || Search/Reading || External || Light/Advanced |
| The following AI functions are '''not''' available in Compass by default. <br/>
| | |- |
| It is an additional module that can be activated by IP7 on request. <br/>
| | | [[EN:Applicant#AI_-_Associated_Companies|AI in the Applicant Search]] || Find affiliated companies || Search || External || Light/Advanced |
| There are additional costs for the AI module. <br/>
| | |- |
| The AI from [https://averbis.com Averbis] has been integrated into Compass. <br/>
| | | [[EN:Result_List#AI|AI Sorting of results lists]] || || Reading || Internal || Light/Advanced |
| | | |- |
| The AI can (in future) be used for various areas in Compass: <br/>
| | | [[EN:DetailView_AI|AI in detail view]] || Simplify reading patents || Reading || External || Light(partial)/Advanced |
| * Folder assignment or automatic classification of patents
| | |- |
| * Automatic sorting out of "not relevant" hits (available soon!)
| | | [[EN:Folders#Folders_and_AI|AI and folders]] || Automatically create/populate folders || Search/Reading || External || Advanced |
| * Sorting of the results list (coming soon!)
| | |- |
| | | | [[EN:AI_Innovator|AI Innovator]] || Query results list (LLM) || Search/Reading || Internal || Advanced |
| The main feature of an AI is that it is trained on the basis of predefined data. <br/>
| | |- |
| The AI should improve or "learn" through constant human correction and renewed training. <br/>
| | | [[EN:AI_Averbis|Averbis AI]] || Folder classification || Reading || Internal || Advanced (paid) |
| AI is therefore always dependent on human performance. <br/> | | |} |
| The quality of the AI is therefore always dependent on this given data or on a human correcting the AI. <br/>
| |
| | |
| | |
| == Folder assignment ==
| |
| | |
| Folder structures are created in Compass to classify patents according to individual criteria. <br/>
| |
| For example, a technology tree can be created to assign patents to specific technologies. <br/>
| |
| With the "Automatic classification" function, the AI can take over this assignment task. <br
| |
| | |
| === Classifier ===
| |
| | |
| [[File:AI_classifier.jpg|800px]] | |
| | |
| Before the AI can assign patents to folders, the AI must learn or be trained based on existing assignments. <br/>
| |
| An existing folder structure with already correctly assigned patents is required.<br/>
| |
| | |
| A "classifier" can then be created.<br/>
| |
| Some settings are defined here, whereby many of the settings affect the training: <br/>
| |
| | |
| *Activate
| |
| Here you can specify whether the classifier can be used or whether regular automatic training takes place.<br/>
| |
| *Name
| |
| | |
| *Training frequency
| |
| This defines how often the classifier is trained automatically. <br/>
| |
| As the contents of the folders change over time, it makes sense to have the classifier trained regularly. <br/>
| |
|
| |
|
| *Max. # Documents per folder
| |
| The maximum number of documents per folder that the AI can use for training. <br/>
| |
| Too few documents per folder will reduce the quality of the training. <br/>
| |
| Too many documents per folder and the training will take a lot of time. <br/>
| |
| <br/> | | <br/> |
| Ideally, the AI should always receive the same number of documents for each folder. <br/>
| |
| For example, if there are a few folders with approx. 500 assignments and many folders with approx. 50 assignments. <br/>
| |
| Then the value should be set to 50 so that training can be balanced and better results can be achieved. <br/>
| |
| The limit value here is a maximum of 1,000 documents. <br/>
| |
| If the number in a folder is higher, 1,000 documents are randomly selected from this folder. <br/>
| |
|
| |
| *Min. # documents required for training
| |
| What is the minimum number of documents that must be in the folder for them to be used in training? <br/>
| |
| If there are only 3 documents in a folder, for example, the AI will not be able to draw any good conclusions from them. <br/>
| |
| A minimum value of 10 documents must be specified. <br/>
| |
| If the number in a folder is smaller, these folders/documents are not used. <br/>
| |
|
| |
| *Min. confidence level
| |
| "Confidence Level" (in per cent) describes how confident the AI is with the assignment. <br/>
| |
| Everything below the minimum value ends up in the "unclassifiable" folder. <br/>
| |
| This is defined later in the automatic classification. <br/>
| |
|
| |
| *Folder
| |
| Here you can define which folders are used for training. <br/>
| |
| The "automatic classification" will later make the assignments in these folders. <br/>
| |
| The AI can then also assign a patent or patent family to multiple folders.<br/>
| |
| A minimum of 3 folders/sub-folders have to be selected. <br/>
| |
| A maximum of 500 folders can be selected. <br/>
| |
|
| |
| *Data
| |
| Here you define which data the AI receives for the training: <br/>
| |
| Title, Summary, Claims, Description, IPC, CPC <br/>
| |
| If Description is selected, the AI receives significantly more data for training. <br/>
| |
| This will also have a corresponding effect on the training time. <br/>
| |
|
| |
| *Status
| |
| The current status of the classifier is displayed here: <br/>
| |
|
| |
| "Training required" <br/>
| |
| After the classifier has been created, training is required. <br/>
| |
|
| |
| "Training" <br/>
| |
| The classifier is currently being trained. <br/>
| |
| During training, automatic classifications that use this classifier cannot be executed. <br/>
| |
|
| |
|
| "Ready" <br/> | | Column: "Supports at" <br/> |
| The classifier is ready for use in an automatic classification. <br/> | | The AI features are designed to support you with patent searches and with reading/evaluation of patents. <br/> |
| This also means that the last training session was carried out successfully. <br/>
| |
|
| |
|
| "Error" <br/> | | Column: "AI Type" <br/> |
| An error has occurred. <br/>
| | IP7 uses a variety of AI systems. <br/> |
| Successful training is a prerequisite for use in automatic classification. <br/>
| | Basically, there are two important distinctions: <br/> |
| | | * Internal |
| *Last training <br/>
| | ** Critical data - will not leave the IP7 system or IP7 server |
| This shows when the last (successful) training session was carried out. <br/>
| | ** Text fields in which, for example, an unpublished invention disclosure can be entered |
| | | * External |
| The current status can be retrieved using the Refresh button: <br/>
| | ** Non-critical data - can be sent to an external AI (interface) |
| [[File:AI_classifier_Status_refresh.jpg|500px]]
| | ** Public patent data, e.g., the claims/description of a published patent |
| | |
| As soon as the classifier has finished training (status "Ready"), it can be used in an automatic classification. <br/>
| |
| | |
| ==== Training statistics ====
| |
| Evaluation of the last training run. <br/>
| |
| [[File:AI_classifier_Status_analysis.jpg|500px]]
| |
| | |
| Further information on the terms Precision, F1Score and Recall: <br/>
| |
| https://en.wikipedia.org/wiki/Precision_and_recall
| |
| | |
| ==== Limitations ====
| |
| In total (across all folders used in the classifier), a maximum of 10,000 patents are used for training. <br/>
| |
| If there are more than 10,000 patents in the folders, these are reduced proportionally. <br/>
| |
| If a folder is then below the minimum number of patents, the selected patents are increased to the minimum number. <br/>
| |
| This means that the limit is not exactly 10,000. <br/>
| |
| | |
| The classifier requires at least 3 folders/subfolders. <br/>
| |
| | |
| === Automatic classification ===
| |
| | |
| [[File:AI_automatic_classification.jpg|800px]]
| |
| | |
| The following settings are available for automatic classification: <br/>
| |
| | |
| *Activate | |
| Here you can specify whether classification is to be carried out automatically. <br/>
| |
| This option can be used to quickly stop an active automatic classification in the event of problems. <br/>
| |
| | |
| *Name | |
| | |
| *Monitored folders | |
| The "Incoming folders" are defined here. <br/>
| |
| All patents in these folders are then classified by the AI. <br/>
| |
| And, logically, all patents that are assigned to these folders in the future. <br/>
| |
| <br/> | | <br/> |
| Not to be confused with the folders of the classifier: <br/>
| | The following general note applies to all AI features: <br/> |
| The folders into which the AI classifies/assigns the patents are defined in the classifier. <br/>
| | Results/outputs generated by AI may be erroneous and/or incomplete. <br/> |
| | | Therefore, they must always be critically examined, questioned, and verified. <br/> |
| *Non-classifiable" folder
| |
| All patents that cannot be classified by the AI are assigned here. <br/>
| |
| | |
| *Classifier
| |
| The previously created classifier is selected here. <br/>
| |
| | |
| *Status
| |
| "idle" - automatic classification is currently not running. <br/>
| |
| "running" - automatic classification is currently running. <br/>
| |
| | |
| *Last checked on
| |
| The system regularly checks whether "new" patents are available for automatic classification. <br/>
| |
| If "new" patents are available, automatic classification is started. <br/>
| |
| The date indicates when a check was last carried out. <br/>
| |
| A check can also be triggered manually using the "Execute" button. <br/>
| |
| | |
| | |
| A classifier can theoretically be used for several automatic classifications. <br/>
| |
| An example of this: <br/>
| |
| | |
| There are several vehicle types that are monitored and should then be assigned to a technology tree: <br/>
| |
| "1, vehicle types" -> "bicycle" and "motorbike" <br/>
| |
| | |
| There is a folder structure or technology tree in which all vehicle types (vehicles with 2 wheels) are to be assigned: <br/>
| |
| "2, two wheel technologies" <br/>
| |
| | |
| However, the hits that the AI cannot classify should be saved separately. <br/>
| |
| For this reason, an automatic classification is created for "bicycle" and "motorbike". <br/>
| |
| In this case, however, the classifier only needs to be created once. <br/>
| |
| | |
| ==== Limitations ====
| |
| A maximum of 5,000 hits/patents can be classified in one run. <br/>
| |
| | |
| If there are more than 5,000, automatic classification is not carried out and is deactivated. <br/>
| |
| If the run is triggered manually, a corresponding warning is displayed. <br/>
| |
| If the run is then started anyway, only up to 5,000 patents are classified. <br/>
| |
| | |
| Patents that have been classified in a run do not have to be removed from the input folder. <br/>
| |
| They are recognised as already classified in the next run and are not classified again. <br/>
| |
| To have one or more patents reclassified, they must be removed from the folder and then reassigned. <br/>
| |
| This reassignment means that they are no longer recognised as already classified. <br/>
| |
