ATLAS ti The Knowledge Workbench This set of

ATLAS. ti The Knowledge Workbench This set of 21 Power. Point transparencies contains information about concepts and use of ATLAS. ti, Please read copyright note on transparency no. 2. Scientific Software Development - Copyright 2001 Thomas Muhr

© Copyright Note The Power. Point transparencies included in this package may be used to support your ATLAS. ti workshops, training sessions & demonstrations. You may alter the transparencies to fit your needs, but please do not remove original copyright notes. If you have any transparencies either self made or created via modification of the existing sheets we will all be happy if you make these available for the public. In no event may the transparencies included in this package be commercially exploited (e. g. , sold) either altered or unaltered without prior written permission by the author, Thomas Muhr, Berlin. Scientific Software Development - Copyright 2001 Thomas Muhr

- The Knowledge Workbench Basics: Advanced Topics: n QDA & ATLAS. ti n n VISE: Visualization, Integration, Serendipity and Exploration Beyond Text: Working with graphics, audio & video materials n Structures: Weaving semantic networks n Users: from Sigmund Freud to Sherlock Holmes n Hypertext: What codes can’t do for you n The main concepts: of Hermeneutic Units, Families and other species n Retrieval: Using Boolean, Semantic and Proximity operators n Strategies: Textual and Conceptual level n Super Codes: Intensional codes or frozen hypotheses? n The user interface: Keep focused on the data n Cooperation: Merging projects Back to the future: The Paper & Pencil look & feel n n Interfaces: ASCII/ANSI, SPSS, HTML, PROLOG, WMF, XML Basic Procedures: Coding, commenting, retrieving, printing, preparing, n n Miscellaneous: Data safety, memo outsourcing, text management, setup, capacities ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 ATLAS. ti

s Visualization s s Integration s s Bundle all relevant data and interpretations into a unique project: the “Hermeneutic Unit” Serendipity s s Use adequate tools for handling complexity and stay focused on the data Make relevant discoveries without searching. . . Exploration s Traverse the “interpretative threads” between data, codes, and memos ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 4 Basic Principles: VISE

Marketing Research Medicine Public Health Literature Education Social Sciences & Humanities Applications Libraries & Archives Theology Astronomy Art Planning Criminology Urban Development ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Areas of Application

ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Textinterpretation as Text-to-Text Compile the primary documents: Texts, Graphics, Audio, Video Open up a “Context of Discovery” to explore the data and add structure Result: another text, diagrams, a WWWdocument ?

Hermeneutic Unit Super Codes Families Networks Code family uses ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 A HU’s Abstractional Layers uses causes isa causes Codes indicated-by Quotations supports contained-in Primary documents indicated-by contained-in

Graphical list of contents ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Graphical Primary Documents Display comments for image sections with a mouseclick

TIF, JPG, Kodak PCD, SUN Raster. . . XML 1. 0 Currently memos and codes Not supported Exported Im- & Export HTML WMF BMP PCD RTF ANSI Will be supported in 5. 0 APN SPSS Presentation Representation ATLAS. ti® SGML ASCII Standardization Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Industry Standards supported by ATLAS. ti 4. 2

Text can be dragged from Win. Word or any other text processors (capable of OLE 2 drag & drop) into ATLAS/ti. Text import is also available via Copy & Paste. A mouse click displays the new memos’ text. Dropped into a Network View, a new memo is automatically created from the text ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Inter-Application Data Exchange

ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 HTML Code Generator The conversion of Hermeneutic Units into HTML code enables new ways of structured publishing. Research teams can quickly exchange ideas and complete projects world wide.

- ATLAS. ti’s main work space Main menu Primary Document area Main toolbar Dropdown fields for Primary Docs, Quotations, Codes and Memos Margin area Selected Quotation Splitter bar to resize panes Context menu Detached code list ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 The HU Editor

Code A While codes describe similarity of the coded segments, it is hard to represent relations (beside the equivalence relation) between individual segments. Only direct links (“hyper-links”) between segments enable the representation of such local knowledge. Q 1 contradicts If one would establish a link between the codes in the example to emulate a hyper link, we would have to assume that these codes do not refer to any other segments, but are used as labels for individual segments: a clear “misuse” of codes. . ATLAS/ti supports named links between data segments. supports Q 2 Code B Q 3 ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Hypertext - what codes cannot do

The Query. Tool retrieves data segments by their attached codes using Boolean, proximity and semantic operators. Queries are entered in RPN calculator style. Boolean operators Semantic operators OR XOR AND NOT Term stack SUB UP SIB WITHIN ENCLOSES OVERLAPPED_BY OVERLAPS FOLLOWS Proximity PRECEDES operators COOCCUR Follows/Precedes distance control Families Feedback pane Codes Stack manipulation Clear stack Swap the two topmost elements Push - duplicate topmost element Recalculate results Undo last operation Redo last undone operation Create Super Code Change feedback display mode Results Textbase selection ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 The Query. Tool

Boolean retrieval is purely set based. Elements are assumed to be independent. No property of a retrieved segment other than being coded with codes A, B, . . X is taken into account. Overselectivity: AND (A 1, A 2, . . . , An) fails even with n-1 matching terms. Underspecified: OR (A 1, A 2, . . . , An) succeeds with everything from 1 to n matching terms. A segment coded with only one code is treated equal to one coded with all of them. A or B not (A or B) A xor B Q 1 Q 4 A Q 3 B Q 5 Q 2 A and not B A and B not A and B Document universe: Q 1, . . . , Q 5 Query examples: A -> {Q 1, Q 2, Q 3} B -> {Q 3, Q 4} not A -> {Q 4, Q 5} A or B -> {Q 1, Q 2, Q 3, Q 4} A xor B -> {Q 1, Q 2, Q 4} not (A or B) -> {Q 5} A and not B -> {Q 1, Q 2} A and B -> {Q 3} not A and B -> {Q 4} ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Retrieval Methods I - Boolean Retrieval

Proximity retrieval takes the spatial relations between the retrieved elements into account. A segment can overlap, follow, enclose or simply cooccur with another segment. The semantics were adapted from Allen’s time logic calculus. Primary document P 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Q 1 B overlaps A -> {Q 3, Q 4} A overlapped by B -> {Q 1, Q 2} C overlaps B -> {Q 5} A within C -> {Q 2} A overlaps C -> {Q 3} C follows A -> {Q 5} B overlaps C -> {Q 3, Q 4} etc. Q 4 Q 5 Q 2 B C A In addition to the Boolean conditions described above, the following proximity relations hold: Q 3 Note, that proximity operators are non-commutative: B op A is not the same as A op B Operand input order is significant! ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Retrieval Methods II - Proximity Retrieval

Semantic, or thesaurus-based retrieval takes transitive relations between the terms (codes) into account. Its quality is dependent upon the quality of the semantic network used. While the extension of sub (Positive Attitude) and or (Love, Kindness) is identical for the example below {Q 1, . . , Q 5}, the intension is different. The former query will - unaltered! yield different results with another subterm of Pos. Attitude. The latter query will not ecognize this new fact and has to be reformulated. Example queries using the semantic operator SUB on the terminology network below: sub (Positive Attitude) -> {Q 1, Q 2, Q 3, Q 4, Q 5} sub (Negative Attitude} -> {Q 6, Q 7, Q 8} sub (Attitude) -> {Q 1, . . , Q 8} Q 1 Attitude isa Positive Attitude isa Negative Attitude sibling isa Love Kindness Hatred isa Anger indicated by Q 1 Q 2 Domain level Q 3 Q 4 Q 5 Q 6 Q 7 Q 8 Document level ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Retrieval Methods III - Semantic Retrieval

Boolean, proximity and semantic operators are combined using the “click-language” par excellence: the Reverse Polish Notation (RPN) by Lukasiewicz 1. RPN is a parenthesis-free postfix language: operands first, then the operators. The main ingredience of the RPN query processor is the Stack, a data structure, that is very similar to a pile of plates: It can only be accessed from the top: new plates are put on the pile, plates can only be removed from the top. Example: “All quotations coded with ‘Positive Attitude’ and any of its sub codes but not with ‘Kindness’” in formal infix notation: SUB Pos. Attitude AND NOT Kindness ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Query. Tool: Building Queries Old HP 29 C RPN calculator Number of arguments Step: 1 Enter: Pos. Att. 2 SUB (1) 3 Kindness 4 NOT (1) 5 AND (2) Stack: Pos. Att. - SUB(Pos. Att) - Kindness SUB(Pos. Att) NOT(Kindness) SUB(Pos. Att) AND(NOT(Kindness), SUB(Pos. Att)) - Result: {} {Q 1, . . . , Q 5} {Q 3, Q 4, Q 5} {Q 1, Q 2, Q 6, Q 7, Q 8} {Q 1, Q 2} Note, how every operator takes (“pops”) its appropriate number of arguments from the stack and “pushes” the resulting term back on the stack. Every entry, operand or operator generates a result. No “syntactic sugar” is needed as in “infix” notations (eg. parentheses). 1 Born 1878 in Lvov (now Ukraine), died 1956 in Dublin, Ireland. Polish Minister of Education in 1919 and professor at Warsaw University from 1920 to 1939)

A well constructed, non-trivial query is often the result of a considerable amount of work and ways to make a query reusable are needed: Super Codes are also an important tool for theory construction as they capture hypotheses for repeated validation against the data. Normal codes store direct quotation references, super codes store queries. Although the visible “clicking behavior” of a super code resembles that of normal codes, there is a considerable difference of “how” each generates its references: Normal codes deliver their quotation references. The result changes only by explicitely assigning new or removing existing references. Query X Unlike other approaches that store the “extension” (the result set) of a query, super codes store the queries’ “intension”. Super codes are “first class” objects and can be used in queries (and in other super codes). Super codes recalculate the stored query “when-needed” and deliver the result. When any of the conditions of the query change, the super codes result list changes as well - without any changes to the latter. ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 The Query. Tool: Super Codes

Merging projects is mandatory for the support of teams working on separate data and/or different code sets. A number of stock merge strategies permits efficient control over the resulting project. Strategies can be freely adapted to fit specific needs. Team A‘s combined project All teams‘ combined project Team B‘s combined project ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Team Work - Merging Projects I

Merging proceeds as subsequent and repeated merging of partial projects into a target project. A merge strategy controls the method of how the different object types (e. g. primary docs, codes) from the source projects migrate into the target project. Examples: A Different data sets, same codes This strategy supports an economic handling of large primary data in a top-down approach. B Same data, different codes By applying this method, different aspects of a theory can be applied to the same data sets. Example (Pi : : = primary documents, Ci : : = codes): HU 1 {P 1, . . , Pn} {C 1, . . , Cm} source project HU 2 {P 1, . . , Pk} {C 1, . . , Cm} target project (before merge) Target project after the merge: Strategie A: HU {P 1, . . , Pn+1, . . , Pn+k} {C 1, . . , Cm} Strategie B: HU {P 1, . . , Pn=k} {C 1, . . , Cm+1, . . , Cm’} ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 Merging Projects II - Strategies

s WYSIWYG - printouts of primary texts plus margin s Media - fine-grained segmentation and coding of video and audio files (incl. MP 3!) s Improved Margin Area s Networks - vector export to drawing software, Word® etc. s Wordcruncher - count word occurrences and calculate type/token ratio. s New reports s Primary Doc Path Mapping Tool s XML - memo and code import & export ATLAS. ti® Scientific Software Development - Copyright 2001 Thomas Muhr© Scientific Software Development, Berlin, 2001 What’s new in ATLAS. ti 4. 2