What is SuperMemo 2008? September 29, 2009

Supermemo 2009 SuperMemo 2008 is a knowledge manager and builder. It helps you build knowledge trough the a mix of enhanced methods for learning and memorization.

Is an integrated environment design so you can read, view or listen any kind of information and following a series of progressive steps, gather bits of information from your fields of interests, and makes sure you remember up the limits of your capacity.

At its heart is a SRS (spaced repetition system) that although looks like other flash card systems, has its own algorithms develop over more than 20 years of research on the field of learning. Not a single other flashcard program offers protection from long delayed repetitions. If you don’t do 100 repetitions for a month you will get 300 items the next day, SuperMemo 2008 knows about human behavior and although it can’t make repetitions for you it helps you deal with delay in your daily repetitions by prioritization of your most important material.

Supermemo is a system integrated modules, so that you can take a big pile of information and processed it according to your current needs but with loosing your creativity purposes. It uses a series of other methods (priority, autopostpone, leaches, etc) developed trough many years to help you with information overflow and remembering what is most important.

Supermemo is better suited for semantic type of knowledge (the kind of knowledge that make you able to remember: What is the name of the president of your country) as opposed to procedural learning (the type that help you remember how to ride a bike). This is the kind of knowledge I more trouble remembering.

Supermemo 2008 has many new teaks over previous version: youtube incremental video has been a great help for getting trough a lot of opencourse clips from great academic colleges as well as many talk on this video platform. I wish this function is extended to other video formats in future builds of the application (it is continually updated), but truth be told I have more then enough to read, view, and listen and Supermemo 2008 is making it so much easier to deal with all this. This is one of the reasons I’ve been lost from blogging for more then 3 weeks (learning, learning, learning… and enjoying the ride).

I’ll make a complete review on SuperMemo 2008 as soon as possible. A lot of enhancements are so great they should have been included from the start… I’m always forgetting supermemo started as a solution for dealing with flashcard learning not the knowledge manager its come to be. Review will be centered in using this new enhancements an not if they are valuable or not, they are!

If this doesn’t fit supermemo’s site description of the program, its because is my personal description for of what it means to me, and probably should mean to you. SuperMemo has a somewhat high learning curve, but is great investment for long term knowledge. If you need fast cram solutions use any kind of flashcards (paper or other software).

When you decide to take the opportunity to enhance your learning with this great tool be sure to go the supermemo store from this link: SuperMemo 2008. That way you’ll buy me a beer!

A better way of Importing MOV files to SuperMemo September 9, 2009

Fastest way to import files into supermemo is by menu options:
File: Import: File & Folders

Previous trouble with MOV files was they didn’t play, because SuperMemo relies on MCI (Multimedia Control Interface).

This Media Control Interface, MCI in short, is an aging high-level API developed by Microsoft and IBM for controlling multimedia peripherals connected to a Microsoft Windows or OS/2 computer, such as CD-ROM players and audio controllers.

For a number of years, the MCI interface has been phased out in favor of the DirectX APIs. Source: Wikipedia

Fastest way to import files is still: File: Import: File & Folders

Just change the .mov extension to .mpg before, that way MCI (SuperMemo) will bypass the trouble with MCI for mov files, then choose the folder with your .mov (quicktime) mpg files and use File: Import: File & Folders

Ok, but then you have to spend hours changing you thousand mov files to mpg manually? Well you could better use AnalogX Extension Changer (download).

This freeware is the most easy to use extension changer. Drop and drag the files you need to change its extension, choose a new one (e.g. “mpg”), hit rename. (drop the complete folder, AnalogX will understand you mean the files not folders).

There you have it easier then typing a Q&A text file. Sometimes you’ll probably prefer text files, but no more trouble importing your MOV files to the sound registry

(ok, I got complex again, not my fault I’m not complicated, Its…. Supermemo fault, for been so much powerful, wright?

SuperMemo 2008 post will come out soon… we must see under the hood before you buy a car don’t we – perhaps not so much this days. I’ll do the testing (under a complete safe environment – complete geekeness I know, but if your going to play with your databank you should!)

Stay tuned.

AnalogX HomePage:  http://www.analogx.com/

Don’t Drill Succefully Recalled Items August 27, 2009

I made explained what procedure I use for drill me when needed (when I make mistakes) during my repetition, you can check out previous post on this procedure here: Drill is only for mistakes

After a couple of topic repetitions today (as promised by Incremental Reading) the theoretical basis for my actions came to me, this is the quote from  the summary of a research article on this issue:

“Supplying the correct answer after an incorrect response not only improved performance during the initial learning session—it also increased final retention by 494%. On the other hand, feedback after correct responses made little difference either immediately or at a delay, regardless of whether the subject was confident in the response” (Pashler, H., Cepeda, N.J., Wixted, J.T., & Rorher,. (2005) When does feedback facilitate learning of words? Journal of Experimental Psychology: Learning, Memory, and Cognition. 31, 3-8.)

After testing the difference is learning versus time spend on repetitions I’m sticking with this police:

Do the Drill if you get more then 1.5 times the desired Forgetting Index (e.g. if your FI = 10% and you get more then 15% then do the drill) but only the mistakes!.

Priority queue issue August 23, 2009

If you like Incremental Reading check out previous post a status report from a newbie in IR

The problem:

Priority queue includes dismissed elements

The experience:

First of all, if you think you read this before see note at the botton of the post.

Browsing trough my collection, I accidentally hit “alt+p” over a dismissed topic (folder in this case). To my surprise it had a high priority position 0.02 % (107 in a 54000 element queue).

To know some elements are never going to be postponed via autopospone is great insurance for high importance material. The only problem though that in this case this is a folder. To keep dismissed elements in the priority queue (even worst high in that queue) is not correct. This elements displace real high priority elements which also affects the autopostpone behavior that is the most effective way to deal with overload and hence with people leaving  supermemo after getting to a huge acumulation of material in their queue. If the autopospone + priority have this design troubles people won’t trust its knowledge to supermemo in the long run.

The priority queue is almost the main reason for using sm2006 but this high priority for dismissed elements simply is plain wrong, and should be even better in newer versions.

The Reason:

Design of a priority queue should  always based on pending material

You are only concern with “to do items”, in supermemo case those are memorized or pending.

Currently the priority percentage is been manage by the total items in the collection, which I feel is a design problem. You have an item with 6% priority in a total of 7000 items of which 50% are dismissed elements, Then in reality its priority is 12%, but of course no one will notice it at first hand.

Dismissed elements by default on any GTD management system goes to the bottom of priorities or even better have no priority at all.

Dismissed elements included in Piority window are useless

In the following scenario:

topic # 67: (dismissed element)

item #24: What is the most common cause of pneumonia? (memorized)

Topic #1542: (dismissed element)

Is not possible to make a decision about priority with regard to previous or folowing element in priority queue. Would you put it before or after the dismissed element, always before anything dismissed, wright?

Autopospone manages high priority material by not postponing them

Auto-postpone is great way for managing overload of material. By using a priority queue things get even better, that is unless the priority queue es flawed.

I consider the information in my collection according to an interval scale:

0-10% high priority

10-50% regular priority

50-100% low priority

Of course all information added to supermemo is always important, but priority has to do with what is most important over other important information.

Setting this interval scaling helps me decide the value of priority assigned to an element when needed, usually a mid interval is chosen inside the desired category

Autopostpone is set to never postpone elements with priorities lower then 10% in my collection.

After a while of including a lot of dissmised items more and more are going to be included with priorities in the 0-10% range leaving many items of high priority aside.

If you got many dismissed elements, and this fill up a lot of spaces in the priority queue, because if you don’t move dismissed elements to the end of the queue they will displaced really high priority elements making this item postponeable and the whole priority + autopostpone algorithm useless.

Proposed Behavior:

• Priority queue should only consider Memorized and Pending elements
• Priority total should only refer to this elements
• Dismissed elements should regain a new priority when added to the memorized or pending material, because of the change in its condition.

Current Work Around:

In my continue motto: “Don’t complain, fix it!”. Here is what I’ve been doing to deal with the problem.

Every once and a while (e.g. monthly or after adding a lot of new elements)

1. Open View: Dismissed
2. Undismiss
3. Dismiss this subset.

Dismissed elements are always put at the lowest priority possible (I previously though this was flawed) In this sense steps needed are less and simpler. Previous steps are bellow in gray.

1. Open complete collection in browser
2. Selected child dismissed
3. Undismiss
4. Change priority
5. Dismiss again the same subset.

Notes: (3) is necessary otherwise browser won’t accept changing priorities in dismissed elements. (4) Use a interval big enough to fit all dismissed items. (5) Don’t ever forget this step, else medicine will be worst then the disease.

I do this is by using autohotkey macros, but not plain keyboard will do it with ease.

This an updated post for problems on priority queue for dismissed elements

Jumpstart Flowchart for Incremental Reading August 17, 2009

Lately we’ve been talking a lot about incremental reading. I was thinking about doing a simplified diagram about using Incremental Reading, when I remember the way I learned first (with out reading the whole help on the subject in the original supermemo website).

Simplified diagram of incremental reading process in SuperMemo

http://www.banaban.net/supermemo/incremental-reading.html

I’m still gathering the worst mistakes made when starting with supermemo, any ideas, feel free to send me  an email. This would certainly make a great post.

gersapa at gmail dot com

Usability not extension determines Flash Card content August 13, 2009

This morning I read a post about discovering that flashcards do work and they are a great way to study, the most important aspect of the post was that the author preferred to write down more complex Q&A than the typical Term-Definition flashcard as used by many online flashcards. The issue of writing them on paper deserves a post of its own, if you are interested in Supermemo, you’re probably already convinced paper can’t match digital anyway (might supplement it though).

Many argue that that flashcards should be very simple (e.a. “flash”) they mostly refer to the issue of extension or wordiness. My own experience, although doesn’t oppose completely to this way of thinking, shows that many simple Q&A questions don’t necessarily need to be short in extension.

The type of information included in my flashcards has suffered constant change, you could probably call it evolving,  trough the simple clozed sentences that later on I had no clue which word was missing and memorizing them felt parrot like repetition, to complex question with a short answer plus further explanations sometimes used now.

Currently my primary way of choosing the amount of information to include in a single flashcard depend mostly on usability, I try to figure out I’ll require to remember the data and what for.

For example if I need to recall what is some “normal” laboratory value like hemoglobin in adults, it makes no sense to include further explanation in the question or answer field. But if values are related by sex, say testosterone values, I would probably make the question emphasize this by asking: What are the serum testosterone values for  (male, female)?. In this specific case I don’t want to ever make a mistake by remembering the female value when a male value is needed, hence remembering this two values together makes more sense than learning them separately.

This seems to contradict the simplicity principle of supermemo, from my point of view though simplicity means as simple as possible. Some knowlegde nuggets need more space for storage in  my external hypocampus thats all.

An example of item with short answer plus explanation:

Which is better Interleaving or Blocking for study?[...]

#Title: Will That Be on the Test?
#Source: Spacing Effect-., Memory-., learning-.,-
#Link: http://www.psychologicalscience.org/observer/getArticle.cfm?id=2425
#Element: 97173: [24/06/2009 04:52 p.m.] Will That Be on the Test?.. Ebbinhaus’s Spacing Effect …
#Article: 96269: Will That Be on the Test?

Answer:

Interleaving is better then blocking

Best is “interleaving” practice items, as opposed to “blocking” them en masse. Example: redesigning syllabi to include short reviews of previous lessons at the end of each class.

How autopostpone affects the learning process. August 9, 2009

• Autopostpone should always be on – with conservative options set.

• Using autopostpone intermittently will make huge deleterious effects on your collection.

Faithful to the GTD management I’ve been trying to develop since a couple of months, a recurring theme about postponing and management of overload came to my focus of interest again.

I’ve always mistrusted leaving my learning processes on absolute autopilot by auto-postponing them, hence I’ve written about this issue before. Today I have a new understanding about what to do and not to ever do when dealing with auto-postpone (postpone for individual branches or subsets is not taken into account in this post).

In order to acknowledge the influence of auto-postpone in the items intervals by supermemo and excel table was constructed. At first I wanted to see a whole lot of combinations of parameters. But for simplicity sake, a single example table is showed  here.

AutoPostpone intervals by number of postpones

The individual intervals and delay factor will of course be different if other parameters are set, in any case I’ll be showing the intervals used in my collection.

Rows show in dark brown the intervals set by supermemo in each postpone (columns), light brown shows the degree of delay (used/optimal interval).

A delay of 2.0 means that the interval set is double of the ideal, 3.0 means triple long interval, etc.

Supermemo establishes the recall probability for each item during repetitions, and this recall probability is directly proportional to the degree of delay, but is influenced by the forgetting index, so this makes comparisons of the probability of recall also relative to preffered parameters.

For a Forgetting index of 20%. A delay of about 2.0 has a recall probability of 84%, and 3.0 has 73% probability, and 4.0 has 64%. So a higher index of delay means, you have to go back to square one on the learning process of the items at hand.

Before, I had autpospone turn off, mainly because I didn’t understand how the postpone parameters affected my learning process by modifying the intervals for the outstanding items and topics. This is no longer a limitations, and in fact I would recommend turning on autpostpone and living it that way afterwards.

Using autopostpone for a couple of weeks or months and the turning it of for another interval of time, and going back again to autopostone will make huge deleterous effects on your collection. Let me show you what I mean.

Lets start by setting a postpone factor of 1.1, max delay of 10, max postpones to 8, and starting optimal interval of 4. Autopostpone will set higher intervals if you do manage to make the repetitions until the item is opstponed 5 times or more.

Optimal Interval = 4
1º postponed interval = 5
2º postponed interval = 6
3º postponed interval = 7
4º postponed interval = 8
5º postponed interval = 9
6º postponed interval = 10
7º postponed interval = 12 (Intervals are set to next whole number, ea. 1.2 = 2)
8º postponed interval = 14 (Intervals are set to next whole number)

Here an item will have an interval of 14 on the 8th postpone, then it will not be postpone any further, here will assume that 4 weeks after 14th day, you finally make a repetition on this item taking an interval of 42 days (14+28).

If you turned of the autopostpone after the 2th postpone, but still don’t make a repetition on this item and before the 42th day you turn on the autopostpone, and by then you would normally (like in the first case) make the a repetition on this item, but because of having turned off before autopospone it will get postpone (until 8 postpones are made, in this example) according to the folowing values.

Optimal Interval = 4
1º postponed interval = 5
2º postponed interval = 6
42th day: autopostpone turned on
3º postponed interval = 42+4.2=47
4º postponed interval = 47+4.7=52
5º postponed interval = 52+5.2=58
6º postponed interval = 58+5.8=64
7º postponed interval = 64+6.4=71
8º postponed interval = 71+7.1=79

Then by not using autopostpone all the time, when you finally get to make a repetition on this item, its interval has gone up to a delay of 19.75 (79/4). Having autopostpone set all the time you get delay of 10.25, still high but not as harmful to your probability of recall as when you use autopospone intermitently. When autopostpone is set on and off more then once the effects are even larger.

How on earth did I got intersted in finding out about this postpone issue. Obviously because I did turn on and off autopostpone as I learned how to set the correct parameters (for my own requirements), only to find out it did make things worst. I started to check if there where items which had last repetition dated more then 30 days ago and that had one repetition (recently entered to the learning process, or mistaken on the last repetition) or two (meaning not yet consolidated). To my surprise many items where repeated more then 400 days ago, with only one repetitions done, and most importantly those where valuable data.

Botton line – Don’t mess with autopostopone use it all the time or never.

Mass importing thousands of audio vocabulary files. August 1, 2009

Thanks to constant interest by fellow blogger Jerry Petit on this issue, I’ve come to the conclusion that indeed adding with ease a set of audio files to supermemo should be a almost automatic process. Its hard to believe this can’t be done using supermemo components. Or has anyone figure out a way?

I’ve added recently a german collection (language related, not about me) from the book2 series which I find quite. Now I’m learning german vocabulary and phrases with out even noticing. So Jerry is positive on this one, we must have and easy way.

Jerry wanted to use html to import Q&A pair. There are many methods for using multimedia by html, I prefer using <embed> gives you many options, and works on almost any browser.

The other portion of the solutions to Jerry’s questioning was about something I also needed in the past. In order to trace the question banks I’m studying (which are printed type, little books, many pages, a waste of time to scan and OCR them) I had to import almost 17,000 references into my collection (no I don’t pretend to finished them… this year). The point is I also had the problem of importing a lot of Q&A in which the main point is keeping track of a series of items, and don’t even have to use the answer field  (the answers are on the question banks, copying them makes no sense). For those who believe that OCR is a great option, in this special case is not, not all this info is needed in my career but could come up in the test, so my option is a matter of efficiency only.

Back to the importing issue.

The way I’ve come up to is using Excel to make  my collection first, then used notepad ++ to take out those nasty tabs that come up when copying from excel to notepad ++. I use excel to do this mainly because is accessible almost anywhere, and also because I’m not good at scripting and less then a newbie in python. Excel does the job of creating a series in a columns and that’s exactly what I need.

Finally I end up with a series of Q&A that look somewhat like the following:

a) If I would like to have the audio Autoplay by default

Q: Question_text<BR><EMBED src=file:///D:/MP3%20songs/0001.mp3>

A: Dummy_text

Q: Question_text<BR><EMBED src=file:///D:/MP3%20songs/0002.mp3>

A: Dummy_text

b) If I would like to manually play the audiofile

Q: Question_text<BR><EMBED src=file:///D:/MP3%20songs/0001.mp3 autoplay=”false”>

A: Dummy_text

Q: Question_text<BR><EMBED src=file:///D:/MP3%20songs/0002.mp3 autoplay=”false”>

A: Dummy_text

Notes:

* <BR> insert new line after the question text

*  “D:/MP3%20songs/0002.mp3″  is just the path to the file in your hard disk, but uses a direct slash and not backslash.

* %20 is used when ever there is a space in you file or folder name.

* Question_text is not needed, but is a good option, perhaps using the corresponding file name would be handy, in order to further purge the collection. But I would probably include the content of the audio is available.

* Leaving the answer portion blank is not an option, else importing Q&A text file doesn’t work. I use only a hyphen when I need this.

After this is done, you will import as any other Q&A file into supermemo:

File menu: import: Q&A text

There is an option to import any kind of files into supermemo without the use of html. Including them by importing option have certain advantages and caveats from my point of view.  Importing this way includes the files as topics, and included a lot of text I don’t like. Of course you can convert this elements from topic to items, still they’re look are not much appealing to me.

So linking mp3 files or any other multimedia type trough hmtl proves a better option in most cases, and as most of us use SuperMemo on a single computer the counter part of absolute linking trough html are not much inconvenient.

Advice:

Links trough html code in supermemo can’t be relative so if you move the source folder where you put your mp3 files, you will have to change the links is your SuperMemo collection, this can also be easily accomplish though.

This has nothing to do with html it self. In SuperMemo each element  is saved in a single file inside a folder that continuously changes as the size of the collection increase, by using relative links you’ll end up with non functional items (or topics).

Linking by using html is easy indeed but does require some work around. If you will like not only to link multimedia files but include then into the collection this can also be done, albeit using other methods.

Wonderful ideas about new uses of SuperMemo on your site Jerry.

Ok, time to train a bit, else my own mother will bet me up at upcoming half-marathon (no I’m not that young, is she who won’t age further, LOL). See ya.

Incremental Reading advice required July 20, 2009

I’m a fan of using incremental reading for reading non vital articles, the kind of articles in the some day maybe list on a GTD system. As a whole it a wonderfull principle, that does make reading more entertained, but if it is more productive than traditional reading or not I’m not quite sure.

I believe the issue of incremental reading has potential for life long learning, once we have a tool to do it. Even better now that thanks to contribution of ideas about multiple color formating by reader Marcin Rybacki multiple ideas can be easily differentiated by themes inside a single article.

SQ3R (Survey, questioning, reading, reviewing, repeating) technique is a good way to ensure that the global idea of text read is acknowledge and as well to learn its content. If it also meant a systematical way to review knowledge afterwards probably this post would not come trough.

A little step at the time, the principles that make incremental reading a better reading technique are being collected in my SuperMemo database of ideas, but one head thinking is very prone to misdirection an overspecialization.

For those of you that use the incremental reading techniques, what are the best principles you would recommend to improved it? Do you believe incremental reading can be used to processes other type of articles, besides the “some day maybe type”?.

For those that don’t use the technique, but have heard about it, which are the reasons you departed from applying it your learning processes?

Very unhappy with postpone. June 27, 2009

In the beginning it was all peace and calm. But as items started to fill my outstanding repetitions, postpone was going to come and help. At least that’s what Dr. Wozniak announced. Until this very momment I have been unable to configure to my best benefit this postpone criteria, somehow I know you can’t make this program to be the more difficult possible software, but still many times I feel Dr. Wozniak just doesn’t understand common human mortals and thinks every single help page in the supermemo site is clearly stated what and what to do in order to accomplish what we want.

Does anybody use postone? even better anyone has configure postpone manually insted of the default values? What are the criteria you use in postpone?

Resetting Previously Postponed Branches May 30, 2009

A couple of months ago I had to organize my postponing so that only certain folders (categories) would be postponed. When you get into the last tab “adjust” in the “Postpone outstanding elements window” in order to see what branches you have previously postponed, there isn’t any way to reset previously postponed branches. This is the question I asked on SuperMemoPedia Wiki.

I have several branches on my collection which have their own postpone
criteria, I can see they have been set before when I go to “Postpone Outstading
Elements” windows, Tab:Adjust, after clicking on “List”. This was not an issue
at first but currently, is frustrating to keep up with which brances have been
postpone before or which hasn’t. I want to reset some of those branches showed
in “List”, so that I can improve the postpone criteria by ignoring some
branches, but other that are in that list. If I choose ignore, all the branches
it it will be ignored, and will maybe I wanted that before, now it is not in my
priority for the automatic postone algorithm.

What is the best way to reset those branches showed on “List”, so they are do
not have individual postpone criteria?

It really did not an answer the question but never the less I learned something important:

“The problem is not with SuperMemo, or with postpone, or priority queue.
The problem is with the appetite for learning more and more. Overload
will always produce the need for postpone, drop in retention,
prioritization problems. Priority queue is currently the most
convenient way of deciding what gets least neglected in learning.” Source: SuperMemopedia

So if you have a great deal of overload then the system will ultimately fail. I’m ok with that. Because of voracious hunger for learning, and need for my profession. But using prioritization has lead to neglecting a whole 4,000 items about items that, although necessary, where constantly postponed because I only relied on priority.

I feel like the postpone, prioritization and forgetting index features are not to easy to set up, but they should at least be easy to take back. And also think setting this options apart give you more control on the items you are willing to postpone or even forget. Setting priority each time I find something interesting is not very effective for all items, as we tend to think that any present item is more important than past learned information, so there is also a priority bias (something described on the SuperMemo site).

Today I’m exactly in the same problem as before, how do you reset previously postponed branches?

Shortest answer:

• Close SuperMemo
• Go into the collection folder (the one name the same as you kno file)
• Search for postpone.ini file and deleted it.

Of course any previous postpone setting will be lost, but that’s exactly what I wanted today.

If you have a lot of postponed branches already set correctly and wan’t to change the postpone criteria for a specific branch has other procedures. That should make a follow up on this issue.

Unconscious Semantic Memorization May 12, 2009

Happens to me every once and a while, I go trough an item, get an error (4 or 5 on SM), this item shoud be reviewed and learned in the drill phase, but can’t ever make it. Dr. Wosniak (SM creator) says the drill is not obligatory at all, and just making the repetitions should suffice, but leeches come repetitively and I find my self hopeless. Will this nugett of information ever get into my hippocampus?

This situation can get very obnoxious. Then, suddenly, It sticks to my memory. I not talking of course of using mnemonics, or other related memory tecnnique, I ‘m just talking pure wrote memory, but in this case “unvoluntary wrote memory”. Neuroscience research has found that even those who have anmesia (with the semantic memory affected) can recall information with out councious effort (they can’t recall when they learned the information in the first place).

Botton line, just go trough the repetitions, if you can find any suitable mnemonic technique to help you out with those leeches, maybe in the interim something comes up to your help, and stop worring about how to learn it, if information is repeated it will get into your memory, voluntary or with out any concious intervention.

I currently working out my note taking and knowledge management procedures, and algorithms, until progress in this issue is done references to the ideas expressed in this blog will be delayed. For now, please forgieve my forgetfulness.