Evidence Tracker (ET) is a tool designed to make the life of scorers easier, by make scoring more efficient and allowing the team to easily track what has and hasn't been scored. At it's core Evidence Tracker is a database of content pulled from the web, that could potentially be of use to scorers. This content is called Potential Evidence Items (PEIs). The PEIs come from numerous sources including; news media, newsletters, X, LinkedIn, company websites and government consultation sites.
The hope is that ET will come to replace the Brian deep-dive and weekly update processes, as well as offer additional functionality and a better user experience Evidence Tracker's features:
A interface that allows scorers to manage incoming PEIs by scoring, dismissing or reassigning evidence items as they see fit
Data is updated on a daily basis
Uses data on who scores which entities to automatically assign evidence items to the correct scorers
Allows the scoring team to have much more direct control over the data gathering process. This is done by letting the team edit; entity - scorer assignments, search terms, entity names and sources to pull data from (Source instances)
ET will be a hub for future scoring improvements, such as AI tools for finding the most relevant evidence items
The best way to run deep dive on a specific entity is to go to the assignment list, find the assignment with that entity (You can always create it if it is not there) and click on "Deep dive results". This will take you to the PEI list with filters set to PEIs connected to that entity that have not been scored or dismissed for the given project you are assigned to that entity for.
To get your weekly updates list, go to the "Weekly updates" under the quick links drop down. This will show you all PEIs assigned to you, from the last two weeks that have not been scored or dismissed for the project you have the most assignments for. Note: If you are scoring for a project that is not your main project, it will be worth changing the dismissed for and scored for projects to the ones you are currently working on. This is so that you don't miss evidence items that have been scored or dismissed for a different project.
From here you can work through the evidence items by viewing them online using the "view online button" and then either dismissing them or scoring them. To dismiss them, you can use the quick dismiss button, but make sure to only dismiss it for a project if you are sure they are not relevant to that project. And only click dismiss for all projects if you are sure it will not be relevant to all projects. If the evidence item is relevant to the project but does not relate to the entity you are scoring, click "view {evidence_uid}", then click "Remove Entity Mention" to remove your entity and "Assignment options to remove yourself". There is also a button under the actions drop down that will do both these things for you.
To score the PEI click on the "view {evidence_uid}" button and score the evidence item. This will add a link under the scored field. ET will then take you to the evidence item created, once here make sure you upload the PDF.
As you dismiss evidence items, they will be removed from the list without having to reload the page. However, once you get to the bottom of the page, you will have to click reload to get the next ten.
In Evidence Tracker assignments are the main way incoming potential evidence is assigned to scorers. These mirror the lead scorer assignments created in LobbyMap and are updated on a daily basis. So if you want to change your official lead scorer, you should do it there. In addition, Evidence Tracker allows for the creation of assignments straight from Evidence Tracker and the creation of multiple assignments per an entity and project. These will be stored with the lead scorer field = "No" and can be used to share out the scoring of an entity if you like.
Entity Names are the names associated with an entity and used to look for the entity in PEIs. Any PEIs that come from sources run by an entity will automatically be connected to that entity (for example, web pages from a website run by an entity). In addition, all incoming PEIs are searched for mentions of entity names (Excluding PEIs from D1). Because of this, entity names can be used to find matches for alternative names entities might have, as well as any subsidiaries of the entity. There may also be cases where an entity names leads to too many mismatches. For example, many PEIs have a button stating "Share on Facebook" or something to that effect, so "Facebook" is a bad name to search for in ET, however "Meta" would work.
In ET, you can create entity names and run searches for them from the entity profiles. It may take a while for a search of an entity name to run, so please only press the search button once and wait for the search to finish. You should see the results number increase when you refresh the page.
Source instances are the individual sources that will be searched i.e. an individual domain, news sites, consultations etc. These sources instances are read straight from the ET database by the data collection code, so any new instance added should have its data appear in the ET database within a day or two. Source instances can be connected to entities, in cases where the entity runs the source, i.e a corporate website. In addition, source instances can also be connected to a scorer, this is for cases where that scorer may want to see all relevant outputs from the source even if it is connected to an entity that they do not score i.e. consultation responses.
Source instances can be created by going to the Source Internet the source instance will belong to. Note: Please do not create source instances if you are not sure of what you are doing. Please speak to Jake if you have any questions.
For some internet sources (mainly D4 sources), source instances (consultations in this case) are pulled from the internet and uploaded automatically. This is so the team can track consultations and decide which ones they would like scraped. All consultations automatically upload, have their stage set to "draft" and will not be scraped. If you want a consultation scraped, then click Post on the change stage button, if you are not interested in the consultation change its stage to 'Dismiss'. The 'Dismiss' stage is functionally the same as draft, but is mainly for checking what source instances have been checked already. There is also a filter function on that list, if you set it to stage = "Draft" and Internet source = "Whatever source you are interested in", then you can work through the source instances dismissing or posting them as you go.
Terms are used by ET to determine if an PEI belongs to a project or not. PEIs will only be assigned to a scorer if it can be connected to an entity they score and it mentions a term from the project they are assigned to for that entity. Hence terms are very important for filtering through PEIs. Terms can be edited or created if deemed necessary. Terms are project and language specific. ET tries to determine the language of incoming PEIs and only checks them for terms of the same language. The "Used in web search" option determines if the term is used in the Google search the D1 data gatherer runs to find D1 results. These terms don't need to be super comprehensive but should just point the Google results in the right direction.