Wiki source code of The AISOP recipe

Last modified by Paul Libbrecht on 2025/06/15 23:32
Show last authors
1 (% class="jumbotron" %)
2 (((
3 (% class="container" %)
4 (((
5 = The AISOP recipe =
6
7 The AISOP webapp is a service built as the result of various training and configurations.
8 This recipe explains how to extract the content fragments, annotate them, and create model trained on it. This will let us create a pipeline and a seminar on which we can analyse portfolios.
9 )))
10 )))
11
12
13 == Basic Terms ==
14
15 The context of the AISOP-web-app usage is that of a course at learning institution which typically has fixed students and fixed contents. A course can contain multiple courses or modules.
16
17 * **AISOP Web-app:** The nodeJS server that interfaces with the portfolio-composing system.
18 * **Portfolio:** the content written by a student in order to represent his or her progress, learning and knowledge using a textual and graphical form. Generally expressed in HTML, can be embedded in various web-pages.
19 * **Course-contents:** The set of slides, their annotations, the videos and handouts that normally read by students and teachers.
20 * **Analysis:** The set of programmes that recognize and measure the contents of a portfolio. Often also the name of the resulting interactive presentation (which can feature summaries or enriched portfolio views).
21 * **Composition Platform:** A space where the portfolio is written. Normally a web-space. In AISOP we have focussed on the classical e-portfol;io composition platform Mahara (a PHP server).
22
23 ----
24
25 == 1) Data Preparation ==
26
27 === 1.1: Make a Concept Map ===
28
29 Employing tools such as CMapTools, create a graphical concept map that represents the topics of the course. This concept map can be familiar with the teachers and learners of this course as a way to show the paths through the content.
30
31 From the concept map, extract a .cxl file which carries the same information and will be presented on the web-page.
32
33 From the concept map, also extract a hierarchy of topics, assuming there is more than (approx) 10 topics in the map. The hierarchy should be a text file with a label per line and the label indented to the right in case of children relation as in the following example:
34
35 >Algorithmization
36 > Flow Charts
37 > Programming
38 > Programming Paradigm
39 > Imperative Programming
40 >Data-Structure
41 > ....
42 >Operating System
43 > ....
44
45 We'll name this file labels-all-depths.txt. From this text file, extract a text file with only the top labels (in the extract above only Algorithmization, Data-Structure and Operating System), named labels-depth1.txt.
46
47 === 1.2: Extract Text of the Course Content ===
48
49 In order for the topic recognition to work, a model needs to be trained that will recognize the words used by the students to denote a part or another of the course. This allows to create relations between the concepts of the course and the paragraphs of the portfolio and offer these in the interactive dashboards. The training is the result of annotating fragments of texts which, first, need to be extracted from their media, be them PDF files, PowerPoint slides, scanned texts or student works. These texts will not be shared so that even protected material or even personal-information carrying texts can be used.
50
51 Practically:
52
53 * Make all documents accessible for you to open and browse (e.g. download them or get the authorized accesses)
54 * Install and launch the [[clipboard extractor>>https://gitlab.com/aisop/aisop-hacking/-/tree/main/aisop-clipboard-extractor?ref_type=heads]] which will gather the fragments in a text file
55 * Go through all contents and copy each fragment. A fragment is expected to be the size of a paragraph so this is what you should copy.
56 * The extractor should have copied all the fragments in one file. Which we shall call extraction.json.
57 * The least amount of content to be extracted is the complete set of slides and their comments. We recommend to use past students' e-portfolios too. We had rather good experience with about 1000 fragments for a course.
58 * If interrupted, the process may create several JSON files. You can combine them using the [[merge-tool>>https://gitlab.com/aisop/aisop-hacking/-/tree/main/merge-json-files]].
59
60
61 ==== 1.2.1: Extract Text from PDF or PNG (PDF → PNG → Text) ====
62
63 Text extraction from PDFs is sometimes faulty. Additionally, many PDFs contain images. To capture this text, Tesseract can be used. A brief explanation of how to use it is provided here.
64
65 **Prerequisites**
66
67 ~1. Tesseract must be installed
68
69 (% class="box" %)
70 (((
71 ##tesseract ~-~-version##
72 )))
73
74 2. Poppler must be installed
75
76 (% class="box" %)
77 (((
78 ##brew install poppler##
79 )))
80
81 **Code**
82
83 (% class="box" %)
84 (((
85 ##for pdf in *.pdf; do
86 # Extract the base name of the PDF without the extension
87 basename="${pdf%.pdf}"
88 \\# Convert PDF to PNGs
89 pdftoppm -png "$pdf" "$basename"
90 \\# Create a text file with the same name as the PDF.
91 for png in "$basename"-*.png; do
92 tesseract "$png" stdout -l deu ~-~-oem 1 | tr '~\~\n' ' ' | sed 's/  */ /g' >> "$basename.txt"
93 echo -e "~\~\n~\~\n~-~--~\~\n~\~\n" >> "$basename.txt"
94 done
95 done
96 rm *.png##
97 )))
98
99 **Explanation**
100
101 ~1. ##for pdf in *.pdf; do ...; done##
102
103 • Loops through all PDF files in the directory.
104
105 2. ##basename="${pdf%.pdf}"##
106
107 • Extracts the filename of the PDF without the .pdf extension.
108
109 3. ##pdftoppm -png "$pdf" "$basename"##
110
111 • Converts the PDF into PNG images, named in the format BASENAME-1.png, BASENAME-2.png, etc.
112
113 4. ##for png in "$basename"-*.png; do ...; done##
114
115 • Processes only the PNG files generated from the current PDF.
116
117 5. ##tesseract "$png" stdout -l deu ~-~-oem 1##
118
119 • Performs OCR on the PNG file.
120
121 6. ##tr '\n' ' '##
122
123 • Replaces line breaks with spaces.
124
125 7. ##sed 's/  */ /g'##
126
127 • Reduces multiple spaces to a single space.
128
129 8. ##>> "$basename.txt"##
130
131 • Appends the recognized text to a text file with the same name as the PDF.
132
133 9. ##echo -e "\n\n~-~--\n\n" >> "$basename.txt"##
134
135 • Adds a separator line (~-~--) after each page.
136
137 **Result**
138
139 • A separate text file is created for each PDF, e.g.:
140
141 • file1.txt for file1.pdf
142
143 • file2.txt for file2.pdf
144
145 • The OCR results of all pages from the respective PDF are written into this text file.
146
147 • Each page is separated by a separator line (~-~--).
148
149 • Temporary PNG files are deleted at the end.
150
151 === 1.3: Annotate Text Fragments ===
152
153 It is time to endow the fragments with topics so that we can recognize students' paragraphs' topics. In AISOP, we have used the (commercial) [[prodigy>>https://prodi.gy/]] for this task in two steps which, both, iterate through all fragments to give them topics.
154
155 **The first step: top-level-labels:** This is the simple [["text classifier" recipe>>https://prodi.gy/docs/recipes#textcat]] of prodigy: we can invoke the following command for this: ##prodigy textcat.manual the-course-name-l1 ./fragments.jsonl ~-~-label labels-depth1.txt##  which will offer a web-interface on which each fragment is annotated with the (top-level) label. This web-interface can be left running for several days.
156 Then extract the content into a file: ##prodigy db-out the-course-name-l1 > the-course-name-dbout.jsonl##
157
158 **The second step is the hierarchical annotation** [[custom recipe>>https://gitlab.com/aisop/aisop-nlp/-/tree/main/hierarchical_annotation?ref_type=heads]] (link to become public soon): The same fragments are now annotated with the top-level annotation and all their children. E.g. using the command ##python -m prodigy subcat_annotate_with_top2 the-course-name-l2 \
159 the-course-name-dbout.jsonl labels-all-depths.txt  -F ./subcat_annotate_with_top2.py##.
160
161 The resulting data-set can be extracted out of prodigy using the db-out recipe, e.g. ##prodigy db-out the-course-name-l2 the-course-name-l2-dbout##.
162
163
164 ----
165
166 == 2) Deployment ==
167
168 Most of the steps below contribute to creating an _[[AISOP-domain>>the-AISOP-domains.WebHome]]_, a directory with all of the subject-specific information. This directory contains runtime information; it can be first composed as a source directory with all the annotated and source documents. The AISOP-domain should be possible to share.
169
170 === 2.1 Train a Series of Recognition Models ===
171
172 The annotation process above will collect all annotations in a single JSONL file. It needs to be split for the different trainings.
173
174 For the level 1, one creates a recognition model by training on the annotated dataset:
175
176 For differentiating among topics of the level 1:
177
178 ##prodigy train the-course-name-l1 ~-~-base-model de_core_news_sm ~-~-lang de ~-~-textcat-multilabel the-course-name --label-stats##
179
180 This outputs a directory called ##the-course-name## with the latest and the best models. Choose one as the chosen model to be copied inside `l1-model` of the domain.
181
182 Then, for each of the L1 topic, you need to separate the annotations (??? Pierre ???) to individual files which you then involve in a training. E.g. here for Error-Correction:
183
184 ##prodigy train the-course-name-l2-Error-Correction ~-~-base-model de_core_news_sm ~-~-lang de ~-~-textcat-multilabel the-course-name-Error-Correction --label-stats##
185
186 Inspecting the output statistics is an effective way to prevent low-quality results for some of the contents.
187
188 Copy each of the produced best models into the domain's `l2-models` directory.
189
190 The two models directories should be put inside `domains` at the root of the AISOP-webapp.
191
192 === 2.2 Create a Pipeline ===
193
194 ... write down the configuration JSON of the pipeline, get inspired [[pipeline-medieninderlehre.json>>https://gitlab.com/aisop/aisop-webapp/-/blob/main/config/couchdb/pipeline-medieninderlehre.json?ref_type=heads]]
195
196 The pipeline is the central configuration information referencing the models and the scripts (found in the `scripts/python` directory). After changing the pipeline and copying it inside the domain, the web-application should be restarted.
197
198 === 2.3 Test ===
199
200 A simple tool to analyze pasted sentences is available using the following schema:
201 ##https://app-url/debug-classifier/model-l1/model-l2/?text=theText##. This allows to verify that the analysis of expected classical sentences performs correctly. The formulation a URLs containing all information allows creators of domains to collect a series of tests that they can perform again after each adjustments (e.g. after a change of hyperparameter of the training or a change of the annotations).
202
203 === 2.4 Create a Seminar and Import Content ===
204
205 Now that the pipeline is effective, we can create, within a group, a new __seminar__ which defines the pipeline and the associated views from within all views accessible to the group administrators.
206
207 The button analyze should the be visible to you. Consider that analyzing all portfolios within a seminar can take time. We have experienced several hours for some cases.
208
209 === 2.4 Interface with the composition platform ===
210
211 The app installation is concluded with the configuration of the service for the Mahara platform: the instructions of the [[aisop-oauth>>https://gitlab.com/aisop/aisop-oauth]] apply here.
212
213 ----
214
215 == 3) Usage ==
216
217 Make sure that the users who are going to be revising others' e-portfolios are group-administrator within Mahara.
218
219 === 3.1 Invite Users ===
220
221 Within a learning management, announce to the expected users of the AISOP-webapp, the availability of the web-app URL. They will login by authorising the app to download from Mahara. The invitation should contain a little description of the expected function.
222
223 The AISOP-webapp is likely able to support the students' writing process.
224
225 === 3.2 Verify Imports and Analyses ===
226
227 Once a sufficient amount of portfolios is available among the views of the seminars. It is possible launch global analyses. These make output synthetic graphs expressing the covergage of each topic. These graphs can be used in classroom to reflect on the course.
228
229 === 3.3 Observe Usage and Reflect on Quality ===
230
231 The AI-based observation can be used and, for each Mahara view, a portfolio-explorer and global-dashboard is available. They both can be the opportunity to observe the e-portfolios quality. They can also reveal classifications' flaws and, most probably, the appearance of topics that don't fit well any of the classification.
232
233
234
235 === 3.4 Gather Enhancements ===
236
237 As a first feedback after the use of the web-app. Multiple refinements of the e-portfolios could be suggested. Several such can be discussed in class.
238
239 Further enhancements can be obtained from the web-app usage: if the proportion of allocated content matches the expectation from the course, if the students' understanding has been suffering for particular contents, or if some content parts are particularly popular.
240
241 Finally, enhancements can be reflected upon for the subsequent issues of the course: Enhancements to the classification, to the contents, to the annotation sources or to the web-app and processes.

Need help?

If you need help with XWiki you can contact: