Quick Start Guide

Please refer to the main assignment page for details or clarification. This is used for both HG2002 and HG8011.

Last Updated: 2018-09-23

Questions? Drop an email to bond (+ @ieee.org). Please make sure the question is not answered here or on the main assignment page.

Getting Started
Annotation
- 2.1 - General Annotation
- 2.2 - What do the numbers and letters mean?
- 2.3 - What if there is no (appropriate) synset to select?

1) Getting Started

Go to the assignment page (This is used for both HG2002 and HG8011)
Scroll down slightly and you will see a list of names. Look for your name.
- [Your name] + [(username {s1, s2, ...})]+ [range of sentences assigned to you]
Click on Start Here beside your name
You will be asked to log in. Click on the "house" ⌂ icon (see image below)
At the login screen:

Your username is indicated in the task page (in the format "s1", "s2", etc, etc)
Your password is Watson (capital "W"). Please change it ASAP!*
* But please, don't change it to something like Holmes...
After logging in, you will be at the dashboard. To change password, click on User Preferences
To change your password: Click on User Preferences. You will see the Change Password screen.

Use this to change your password.
Go back to the Disambiguation Task page, and click once more on "Start Here" beside your name.

This time, you will be logged in

Make sure the username indicated is yours!

2) Annotation

2.1 - General

After you clicked "Start Here" beside your name, you will see the main annotation interface:

The interface is split into two panels:
- The left panel contains the annotation text and tools
- The right panel contains the senses available available to choose from for the selected word
On the annotation half of the screen (left panel):
- The word highlighted in yellow is the currently selected word.
- When you begin, the first word of your first sentence should already be highlighted for you
- Words that are red and underlined are untagged
- Words that are dark green-blue and underlined are tagged
- You will see some sentences before and after your actual sentence (ie, the context)
- Clicking on words in another sentence will shift the context up (or down).
  - Clicking on a word in sentence 55886 will show sentences 55883 – 55889
  - Clicking on a word in sentence 55882 will show sentences 55879 – 55885
- Therefore, as you move sentence by sentence, the context will shift with you.
- You can widen the context by changing the Sentence Context textbox at the bottom of the left panel.
  - Context is inclusive of the main sentence, bothways (as can be seen in the picture above)
- You can change the active/main sentence by changing the Goto sid textbox at the bottom of the left panel.
On the synset half of the screen (right panel):
- A list of available senses, plus the lemmas in the senses, their definitions and examples.
- Each number has a subscripted letter (Example: 01_a).
  - n - Noun
  - v - Verb
  - a - Adjective
  - r - Adverb
  - x - Others
- Right at the top is a bar that shows "All", N, V, A and R. These are used to filter the synset results.
You will be presented with at least one tagging bar for the word. The numbered choices on the tagging bar correspond to the numbered choices for the synsets, as illustrated below:
To tag, simply select the appropriate synset via the tagging toolbar. (What do the numbers and letters mean?)

Your choice will be automatically saved once selected. A notification message will appear at the bottom to indicate it has been saved.

(In the above image, I have selected "1_a" as the synset used to tag the word.)
- Some words can have a lot of senses, and some senses might be quite close to each other in meaning.
- As such, you will need to pick the best meaning
- We certainly welcome thoughtful tagging, but do not overthink!
- Make use of the Comments box. (You'll see some uses for it below)
- What if there is no (appropriate) synset to select?
Repeat the above step for all other words, until you have finished tagging your assigned sentences.
If it is a multi-word expression (MWE), you might be presented with multiple choices (multiple tagging bars):
- Click on the lemma of the other choice(s) (in this case, ground floor) to see the synsets available for it.
- Tag the multi-word expression (Eg: ground floor) with a synset.
- Tag the individual words (Eg: ground and floor) with x, as illustrated below.
  (Do make sure you tag all the individual words with x)
If the word...
- ... is mis-spelled, mis-tokenized, etc, tag it with e
- ... does not appear in any synset, tag it with w, and make a suggestion (See What if there is no (appropriate) synset to select?)
- ... should not be tagged with any of the synsets, tag it with x
- In all cases, write your reasons in the Comments box
- See What do the numbers and letters mean? for more information

2.2 - What do the numbers and letters mean? (↑ Back)

In the tagging bar, you will see some letters as well as a series of numbers (if any). The table below explains what they are:

	Description
1, 2, 3...n	These numbered choices correspond to the numbers of the synsets shown in the right panel. The letters in subscript (Eg: 1_a, 2_n) are the parts of speech
e	Used to indicate an error in corpus/text (spelling, lemmatization, tokenization, etc) Eg: does being mistakenly analysed as doe+s and not do+es in "He does not know" or leaves as leaf, instead of leave: In this case write the correct lemma and synset in the Comments box. Refer to the Detailed Documentation for more cases to use 'e'
x	Used to indicate that there is no need to tag the word Examples include closed class parts-of-speech, such as auxiliaries (Fred is swimming; Fred has swum) (Note: Copula is (to be) and verb has (to have) should still be tagged) Refer to the Detailed Documentation for more cases to use 'x'
w	Used to indicate suggestions or changes to the wordnet. These include suggestions to add/remove lemmas, or to add a new synset. You can also write other comments such as `"I can't really decide between 012345678-x and 012345678-y"` Refer to the Detailed Documentation for more cases to use 'w'
↓ For names (or part of a name), use the tags below ↓
	Note: If the name already exists in the Wordnet as a synset (such as "Holmes"), choose that synset instead.
Org	Organisation Eg: Scotland Yard (Note: Tag both Scotland and Yard individually as org)
Loc	Location Eg: Riding Thorpe Manor; Norfolk (Note: Tag Riding, Thorpe and Manor individually as loc)
Per	Person Eg: Irene; Irene Adler (Note: Tag both Irene and Adler individually as per)
Oth	Any Other proper name Samsung Galaxy S5 (Note: Tag Samsung, Galaxy and S5 individually as oth)
↓ For other types, use the tags below ↓
Dat	Date/Time Eg: 3:13am; 26/01/1788
Num	Number Eg: 12756; 1138; 802701 (Note: If the number already has a synset, tag it with that synset instead)
Year	Year 1984; 2016 (Note: Be sure this is not just a normal Number)

2.3 - What if there is no (appropriate) synset to select? (↑ Back)

In some cases, a word might not have any (or any appropriate) sense/synset to select in the right panel.

This can mean a few things:

The word is not in any of the synsets (yet)
The synset for that concept does not exist at all

In such a case, tag the word as 'w'.

Now, begin by assuming the word (and its lemma) is not in a synset, but the concept of it exists somewhere in the wordnet.

You should then try the following:

Search for orthographical variants (Eg: stir-fry might be stored as stirfry in the wordnet)
- If found, note in the Comments box. (Eg "Stir-fry: Orthographical variant of 'stirfry' in 00326459-v")
Search for synonyms (Eg: You can search for "funny" if there is no synset for, say, comical)
- If found, note this in the Comment box. (Eg "This lemma is a synonym of 012345678-x")
Search for a hypernymous concept and see if its list of hyponyms contains the concept you want
- Or vice-versa: search for a hyponymous concept, and see if the concept you want is a hypernym.
- Also, search for possible co-hyponymous concepts
- If found, note this in the Comment box. (Eg "This lemma is a hypernym/hyponym/co-hyponym of 012345678-x")
Example: Sherlock currently does not belong to any appropriate synset.
- So we search for something similar — a synoynym of sorts — such as Holmes, using the Search box at the bottom of the right panel.
  - We could also search for Sherlock Holmes.
- Synset 09604451-n appears to be what we want. It's a hit! (In other cases, we might have to try a few synonyms before we get an appropriate synset)
- We tag this word with w, if you haven't already done so.
- Next, in the Comments box, we write:
```
	Lemma: Sherlock
	This lemma is a synonym of 09604451-n
				
```

» Refer to the Detailed Documentation for more on what to write in the Comments box

Ok, you tried the above, but still nothing turns up.

It could thus be that the concept does not exist at all in the wordnet. In this case, you can suggest a new synset.

Search for other relations that you can use to describe the proposed new synsets
- antonyms, hyponyms, hypernyms, etc. (See the detailed documentation for more information)
Come up with a definition for the proposed new synset, as well as an example, if possible.
- If you use the definitions found in another resource (Oxford Dictionaries, Wiktionary, etc), please indicate the source.
Write the definition in the Comments box

Here is an example of the format for writing your definitions:

		Lemma: arrow (Singapore English)
		This is a hypernym of 012345678-x
		Def:	"To assign a task to someone. Generally used only if the task is unpleasant or boring."
		Ex:     "They come and arrow me type their document"

Questions? Drop an email to bond (+ @ieee.org)