Jump to content


  • Posts

  • Joined

  • Last visited

  • Days Won


Everything posted by tufelkinder

  1. Does anyone have any experience or advice regarding configuring Nginx as a reverse proxy for Intella Connect? Or any advice regarding any reverse proxy and successfully passing authentication? The login page loads perfectly, but I keep getting an authentication error from Connect no matter what configuration parameters I use: location / { proxy_buffering off; proxy_set_header Host $host; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_set_header Authorization $http_authorization; # this doesn't work proxy_pass_header Authorization; # this doesn't work either proxy_set_header X-Forwarded-User $remote_user; # and this still doesn't work proxy_pass http://intella-host.local:9999/; }
  2. One of the biggest issues I've had to deal with to date is managing duplicates across a case and across a custodian. For example, in a current case the entire company has a policy of emailing their logo as an attachment. I now have 100,000+ copies of this image in the case. However, if one of my reviewers marks an email as Relevant and another users marks a different email as Non-Relevant, but they both have Tag duplicates and Tag families turned on, this creates a situation in which I have a lot of files tagged with both Relevant and Non-Relevant. On the other hand, if I don't tag with duplicates and my search includes something like "Exclude tagged items," then the duplicates of tagged items appear to populate the list after one of the duplicates has been tagged, so the count of untagged items remains confusingly high. For example: Main search: 6,438 documents Deduplicated: 4,011 documents Tagged: ~3,000 documents Remaining untagged items: 2,271 But there should be only ~1,000 untagged documents remaining. I'm not sure what the right answer is, but it seems like there should be some way of de-duplicating the dataset (at the custodian level at least) and then operating on that deduplicated set.
  3. Hi Łukasz, Custodians A couple useful functions that could be achieved from having this as a built-in object type would be managing Custodians as a datapoint/field when exporting/producing data (Not sure tags would help with this?), organization of evidence/data, and reporting and stats. On a recent case, we were asked to provide hit counts for a keyword set across custodians. In Relativity I was able to do this using the Pivot table function -- not a fan of Relativity, but it was nice to be able to get these kinds of stats within a single function. Or, how many hits per custodian have been reviewed and how many are remaining. A lot of times we just use the "has tags" criteria to determine if something's been reviewed, so if we are using tags for custodians then every document will be tagged by default. I think hierarchical tags will be just awesome, and it could definitely help with managing the data, but it seems like the Location facet can get really messy on large cases and it would be more natural to be able to logically organize the data there directly. You and Adam are correct, I think: doing 20 exports would have solved the problem in this case. I honestly didn't think of it until after I'd already produced and the client came back to me for the "missing" duplicate attachments. Text Analytics When I first saw Concept Search in Clearwell it definitely sounded interesting but some of my reviewers have reported that in real world use cases it can be a little "pie in the sky" when put to practical use. The text analytics I'm imagining of would be very concrete and transparent: show me the words most associated with this word. If you think there's value in this, I'll put some more thought into it and come back with better examples and details. Two-factor Authentication Sounds great -- definitely don't want to be the only person interested in it so I'm curious if you get any other response. We are transitioning many of our clients over to two-factor on other services like Google Apps and Box.com and it always shocks me when people don't ask for or seem to expect more security from their eDiscovery platform.
  4. Maybe I'm not using the correct terminology. The client wanted each custodians dataset deduplicated against only that custodian's data. Let me try to describe what actually transpired: I did a single production of data from these custodians and provided it to the client. They came back to me and complained that the duplicate copies of attachments across custodians had not been produced. So, if Custodian James sent an attachment to Custodian Bill, I was producing two emails (one from James' mail and one from Bill's) but only one attachment -- even though the attachment was technically included in the email file -- because Intella was deduplicating the extra copies of the attachment. This was for a relativity load file so that wasn't working for them and I had to do an additional production of all of the duplicate copies of the emails. I'm not sure that this explanation is any clearer...
  5. If I turn on deduplication and then search all locations as you described, won't it de-deuplicate across locations?
  6. Indeed, oops! So the problem with using Location as custodian is that there is no ability to edit sources/locations after they've been added or relate them to each other or organize them. For example: if I receive a hard drive image for a custodian, I can place it in a "custodian" folder and index it. Then, if I want to later add a new source of data, such as a server PST file, or a cell phone image or something, then if I add that to the custodian folder I need to re-index the whole case. If I add it as a separate source I can't associate it with the custodian folder. At least, unless I'm missing a new feature that lets you organize the Location facet, which would be awesome. Secondly, I had 22 custodians with this case and doing 22 separate exports would have been a bit of a pain and probably wouldn't have made the recipient of the productions too happy either.. But yes, that would have been a solution to the problem.
  7. Just to follow-up: the lack of ability to assign subsets of documents to reviewers as batches cost me a project to Relativity, so this would be really beneficial. Plus, I think Intella could build a lot better Batch function than Relativity has. For example, live batches vs. static batches: live batches could be based on a search or some criteria that automatically updates as new evidence is added to the case. So the reviewer can just check for new documents and they will appear when they are available. Also, a cool component would be if documents disappeared in real-time from the bubble and document list as they were reviewed. This way if the user closed the Preview screen and came back to the list, they would be able to see without clicking on anything where they left off and how many more documents remained for them to review. This would also be beneficial when multiple reviewers are working on an overlapping document set. If the they have exclusive searches, as one reviewed a document it could automatically be removed from the other users' review set(s). This would also eliminate the need for individual batch creation and assignment. If all users had access to a pool of documents and the system prevented review documents from being displayed to more than one users, then individual batches aren't as essential, and you can always track back to see stats on each reviewer.
  8. Hierarchical tags sounds awesome! I can't wait to see this in action. Definitely a feature that my users have been requesting for a long time... Two more wish list items for you. 1) Custodians: The ability to organize data by definite Custodians would be really helpful in making data interchange in and out of Intella more consistent and universal across other products. To take it a step further, we recently ran into a major export where the client was requesting Custodian-level deduplication and this posed a challenge for us to accomplish efficiently in Intella. 2) Simple Text Analytics: I've been asking for this for ages, but I wish there was a Word Explorer that would allow you to view/filter/browse word associations much the same way that the social network tool works. I know that the word indexes are there "under the hood," but if we could access them through the UI I think this would make for really powerful "technology-assisted" review. For example, if you search for a keyword the Word Explorer could show you other words that are highly associated with your keyword, or which custodians use the keyword the most, etc. Link this up with a thesaurus and/or dictionary and you could include variants, synonyms, or alternative spellings of the keyword. 3) Two-factor authentication for Connect: since Intella almost by default is hosting extremely sensitive information, it would raise the level of security and increase awareness of the sensitivity of the data for my clients and users to have two-factor auth enabled. Plus, it's a great feature that almost no other platforms out there have in place at this point. Google Authenticator would be perfect for this. Thanks for all your continued good work!
  9. Intella 8 doesn't seem to respect the Indexing setting for Index Content embedded in documents; when I indexed identical material between the two versions with this setting disabled, 1.8b1 generated ~11,000 results, 1.7 generated ~6,500. I would like to see more permissions options (such as: - Review Case and Apply but not Edit/Create Tags - Restrict User to These Tags (the inverse of the current option) -- this could be used to create batching, while batching with the current method would be difficult to manage. The Type and Locations Facets do not load in the Manager, and the Type Facet does not load with any values in the web interface. I'm sure I'll have more feedback shortly, but my first reaction is overwhelmingly positive to the new UI.
  10. I would very much like to test this out!
  11. Oh, like a time range slider, or "all events between these dates" filter... if it could visually update/animate over a certain time range? That would be really cool! You could watch traffic patterns build between two people, for example. Regarding the merging of entities, automatic would be great for sure, but I'm just thinking in cases where you have a clear person of interest, it would helpful to be able to manually group multiple email addresses (or other identifying information) into a single node. All of these features sound excellent! Now if I could just import phone call logs as well... :-)
  12. So I'm definitely loving the new Social Graph feature in 1.7.1. However, there is one feature you could add that would immediately improve its utility for me: • A threshold setting or slider that filters out links between participants if the number of messages exchanged is below a certain number of messages/hits. Longer term suggestions (which I'm sure are already in the works) would include: • Ability to aggregate variant email address information (or various email addresses, phone numbers, etc) into a single entity. • A word graph of search terms; if I search for a term, show me the other words (and/or participants) that are most highly correlated with that term, with the same filter as the above suggestion. • Abilty to save the results of all of the above adjustments, and any visual movements or adjustments that are made. Excited to see this new feature and looking forward to where it will be going in the future!
  13. When exporting eml/email files from the Intella the file is empty. Any ideas what might cause this behavior? Also, there appears to be no way to quickly jump to an ID number. So if someone says, "I'm looking at Item ID 194572," is there a way to quickly get to that document by entering the ID?
  14. The HTTPS guide worked very nicely. Thank you Łukasz. One thing I'm noticing now: when previewing some (all?) emails, long lines are not being wrapped in the text area so I must keep scrolling back and forth in the previewer's contents textbox to read the message. Thanks!
  15. This appears to have worked well, thank you!
  16. Is it possible to reconfigure the Intella connect server to use a different port (like 80, for example) as the admin console port?
  17. I, for one, really dislike the entire Relativity interface and platform. It's ponderously slow, clunky, and non-intuitive for new users. Plus, advanced features are frequently obscure and painful to access, and building complex searches is challenging for the new user, to say the least. As a result, in my experience, the average lawyer only learns as much as necessary to use it as a linear review tool, rarely as a powerful search platform. I've been rolling out Intella Team for over a year to some of the lawyers I work with, and Intella Connect since it came out in Beta. Frankly, training lawyers who have never seen Intella to do powerful searches is substantially easier than training lawyers who have never seen Relativity/dtSearch to use those tools. Typically a 30-45 minute initial training, and then a 15-20 minute follow-up within the first week, and then I have very few questions or ongoing tech support. The appeal of releasing a tool that "looks like" Concordance or Relativity to avoid training them to use something better seems to defeat the whole point of developing a better interface in the first place. That said, philrodo's and Łukasz's comments on a more configurable interface sounds great -- I just get nervous when I hear talk about downgrading the powerful tool that has worked so well for us. :-)
  18. I kind of hacked together a solution if anyone is ever interested. I converted the TIFFs to PDFs and OCRed them, then wrote a Python script to parse the load file and join the pages into a single PDF where they belonged. Worked pretty well, but a little tedious.
  19. I have received a load file with TIFF images and extracted text instead of a native document set. What is the best way to deal with/process these? Thanks in advance...
  20. I don't want to seem too exuberant, but I am really excited about this feature. How do we enable it? I see that there are new fields under Settings, but are their instructions for configuration? Also, I'm really liking the way the facet windows scroll now instead of expanding to full height. Thanks!
  21. Thanks Łukasz, much appreciated. I'm really loving the product and excited about where this is going!
  22. Does the Intella Connect beta expire? And, of course, the follow up being when can we purchase it? :-)
  23. Awesome about SSL. Really looking forward to that! The trouble with the facet scrolling is that when you have very many, you have to scroll down to click Search/Include/Exclude and it scrolls the preview and content area past the top of the screen. Having a tree structure for tags would enable you to organize your tags by overall categories, so for example, in a large investigation you might have topics like this: • Individuals • John Doe • Joe Smith • Jane Doe • Executives • Raymond Reynolds • John Xavier • Ralph Lauren • Violations • IP Theft • Collusion • FCPA • Money Laundering • Locations • New York, NY • London, UK • San Francisco, CA and each one could expand with many different sub-topics throughout the investigation without causing the overall Tags palette to become thoroughly unwieldy. Plus, when you attempt to apply a tag to a new item, you can browse directly to it by overall topic instead of scrolling through a long alphabetized list, which currently puts Ralph Lauren next to San Francisco and Money Laundering next to New York. The next best thing would be to have a Filter text field so that you could quickly narrow down the tags list by typing in all or part of the tag name...
  24. Okay, definitely impressed with Intella Connect for a beta! Two questions: 1) SSL? Is it possible for users to "hack" the http server and add SSL even if Intella doesn't/won't officially support it? 2) Facets: whenever there's a long list (such as a few dozen tags) , scrolling scrolls the entire screen instead of just the facets panel. More generally, it would be great if we could create a proper tree structure for tagging... Thanks! Keep up the good work! -~
  25. It would be great to Beta Connect! I heard Peter talk about it and do a small demo of it in DC a couple of months ago and have been excited to hear more ever since!
  • Create New...